You are on page 1of 52

52 pages including cover

CSI Communications Knowledge Digest for IT Community

` 50/-
ISSN 0970-647X

Volume No. 40 | Issue No. 6 | September 2016   


A Framework for
Segmentation 28

Medical Image Processing:

An Overview 7
Education In India
TECHNICAL TRENDS – A Roadmap to the
Trend in Medical imaging 26 Future 35
Know Your CSI
Executive Committee (2016-17/18) »
President Vice-President Hon. Secretary
Dr. Anirban Basu Mr. Sanjay Mohapatra Prof. A. K. Nayak
309, Ansal Forte, 16/2A, D/204, Kanan Tower, Director, Indian Institute of Business
Rupena Agrahara, Bangalore Patia Square, Bhubaneswar Management, Budh Marg, Patna
Email : Email : Email :

Hon. Treasurer Immd. Past President

Mr. R. K. Vyas Prof. Bipin V. Mehta
70, Sanskrit Nagar Society, Director, School of Computer
Plot No. 3, Sector -14, Rohini, Delhi Studies, Ahmedabad University, Ahmedabad
Email : Email :

Nomination Committee (2016-2017)

Chairman Dr. Santosh Kumar Yadav Mr. Sushant Rath
Mr. Ved Parkash Goel New Delhi SAIL, Ranchi
DRDO, Delhi

Regional Vice-Presidents
Region-I Region-II Region-III
Mr. Shiv Kumar Mr. Devaprasanna Sinha Dr. Vipin Tyagi
National Informatics Centre 73B, Ekdalia Road, Jaypee University of Engineering and
Ministry of Comm. & IT, New Delhi Kolkata Technology, Guna - MP
Email : Email : Email :
Region-IV Region-V Region-VI
Mr. Hari Shankar Mishra Mr. Raju L. Kanchibhotla Dr. Shirish S. Sane
Doranda, Ranchi, Jharkhand Shramik Nagar, Moulali, Vice-Principal, K K Wagh Institute of
Email : Hyderabad, India Engg Education & Research,Nashik,
Email : Email :
Dr. K. Govinda
VIT University, Vellore
Email :

Division Chairpersons
Division-I : Hardware Division-II : Software Division-III : Applications
Prof. M. N. Hoda Prof. P. Kalyanaraman Mr. Ravikiran Mankikar
Director, BVICAM, Rohtak Road VIT University, Vellore Jer Villa, 3rd Road, TPS 3,
New Delhi Email : Santacruz (East), Mumbai
Email : Email :

Division-IV : Communications Division-V : Education and Research

Dr. Durgesh Kumar Mishra Dr. Suresh C. Satapathy
Prof. (CSE) & Director-MIC, SAIT ANITS, Vishakhapatnam
Email :
Email :

an individual. CSI Headquarter :

Samruddhi Venture Park, Unit No. 3, Publications Committee
2 are friends. 4th Floor, MIDC, Andheri (E), Prof. A. K. Saini
Mumbai-400093, Maharashtra, India GGS Indraprastha University
3 is company. Phone : 91-22-29261700 New Delhi
Email :
Fax : 91-22-28302133
more than 3 makes a society. The Email :
arrangement of these elements makes
the letter ‘C’ connoting ‘Computer CSI Education Directorate : CSI Registered Office :
Society of India’. CIT Campus, 4th Cross Road, Taramani, 302, Archana Arcade, 10-3-190,
the space inside the letter ‘C’ connotes Chennai-600 113, Tamilnadu, India St. Johns Road,
an arrow - the feeding-in of information Phone : 91-44-22541102 Secunderabad-500025,
Fax : 91-44-22541103 : 91-44-22542874 Telengana, India
or receiving information from a computer.
Email : Phone : 91-40-27821998
2u u



Chief Editor
DR. A. K. NAYAK Contents
Cover Story
DR. VIPIN TYAGI Medical Image Processing : An Overview 7
Jyothi Shetty and Niranjan N Chiplunkar
Medical Imaging and Processing 10
Published by Dushyant Kumar Singh
MR. SANJAY MOHAPATRA Pathological Brain Detection Systems : A Review 13
Deepak Ranjan Nayak, Ratnakar Dash, Banshidhar Majhi and Pankaj K. Sa
For Computer Society of India
Overview of Medical Image Segmentation in MR Human Head Scans 17
R. Siva Shankar and K. Somasundara
Design, Print and Medical Ultrasound Imaging 20
Dispatch by A. R. Anil and N. K. Ragesh
Medical Image Retrieval using DICOM Format 23
Amol Prakash Bhagat and Mohammad Atique

Technical Trends
Trend in Medical imaging 26
Nityesh Bhatt and Abhishek Agrawal

Research Front
A Framework for Hyperspectral Image Segmentation 28
B. Ravi Teja and K. Venkatarao
Model Based Approach for Effective Diagnosis of MS-Lesions 31
K. Naveen Kumar and Y. Srinivas

Engineering Education In India – A Roadmap to the Future 35
N. J. Rao

Please note:
CSI Communications is published by Computer
Society of India, a non-profit organization.
Views and opinions expressed in the CSI
Communications are those of individual
authors, contributors and advertisers and they
may differ from policies and official statements
of CSI. These should not be construed as legal
or professional advice. The CSI, the publisher, PLUS
the editors and the contributors are not
responsible for any decisions taken by readers
on the basis of these views and opinions. CSI Service Awards 2016 30
Although every care is being taken to ensure
genuineness of the writings in this publication,
CSI Academic Awards 2016 39
CSI Communications does not attest to the
originality of the respective authors’ content. Brain Teaser 41
© 2012 CSI. All rights reserved. CSI Reports 42
Instructors are permitted to photocopy isolated
articles for non-commercial classroom use Student Branches News 45
without fee. For any other copying, reprint or
republication, permission must be obtained
in writing from the Society. Copying for other Printed and Published by Mr. Sanjay Mohapatra on Behalf of Computer Society of India, Printed at G.P. Offset Pvt. Ltd. Unit-
than personal use or internal reference, or of
articles or columns not owned by the Society
81, Plot-14, Marol Co-Op. Industrial Estate, off Andheri Kurla Road, Andheri (East), Mumbai 400059 and Published from
without explicit permission of the Society or Computer Society of India, Samruddhi Venture Park, Unit-3, 4th Floor, Marol Industrial Area, Andheri (East), Mumbai 400 093.
the copyright owner is strictly prohibited. Tel. : 022-2926 1700 • Fax : 022-2830 2133 • Email : Chief Editor: Dr. A. K. Nayak

u 3 u



Dear Fellow CSI Members,

Medical Image Processing deals with the techniques and processes of creating visual
representations of the body for medical analysis. It helps in extracting useful information
from various body parts for diagnosis and treatment. It helps in identification and analysis
of body parts that may not be visible by human eye. It includes the enhancement and
analysis of images captured using various techniques like x-ray, ultrasound, MRI, optical
imaging. Computer algorithms can provide analysis in 2-dimensions or 3-dimensions to
identify patterns and characteristics which are indicative of various ailments.
Medical Image Processing is an interdisciplinary area consisting of Computer science,
Information Technology, Medicine, Artificial Intelligence, Mathematics, Physics, Statistics,
Psychology etc. The recent advances in medical image processing will help in making
significant advances in understanding of life and disease processes, and ability to deliver
quality healthcare.
Keeping in mind the importance of Medical Image Processing in today’s context, the publication committee of Computer Society
of India, selected the theme of CSI Communications (The Knowledge Digest for IT Community) September 2016 issue as “Medical
Image Processing”.
Cover Story contains first article “Medical Image Processing: An Overview”, J. S. Niranjan and N. N. Chiplunkar that explains
various steps in medical image processing Next cover story “Medical Imaging and Processing” by D. K. Singh gives an overview of
medical imaging with the help of an example of brain tumor detection. Next article in this category “Pathological Brain Detection
Systems: A Review” by D. R. Nayak, R. Dash, B. Majhi and P.K. Sa provides an overview of system used to detect abnormalities in
the brain correctly with the help of MRI scanning. In next cover story “Overview of Medical Image Segmentation in MR Human Head
Scans”, R. S. Shankar and K. Somasundara provide an introduction to segmentation technique in medical image processing.
Cover story “Medical Ultrasound Imaging “ by A. R. Anil and N. K. Ragesh discusses the ultrasound imaging and image quality
enhancement. The article “Medical Image Retrieval using DICOM Format” by A. P. Bhagat and M. Atique provides information
about a widely used format in medical imaging called DICOM.
Technical Trends contains “Trend in Medical Imaging” by N. Bhatt and A. Agrawal that provides advantages of digital processing
in medical applications and discusses imaging market in India and world.
Research Front category contains “A Framework for Hyperspectral Image Segmentation” by B. R. Teja and K. Venkatarao that
gives a framework to handle hyperspectral image segmentation. Next article in this category “Model Based Approach for Effective
Diagnosis of MS-Lesions” by K. N. Kumar and Y. Srinivas addresses a methodology for identification of MS lesions from the brain
MRI images.
We are thankful to Prof. N. J. Rao for sharing his article “Engineering Education In India – A Roadmap to the Future” that reviews
the challenges of engineering education and provides a roadmap to tackle these challenges.
This issue also contains Crossword, CSI activity reports from chapters, student branches and Calendar of events.
I am thankful to Prof. A. K. Saini, Chair-Publication Committee and entire ExecCom, in particular to Prof. A. K. Nayak and Prof. M.
N. Hoda for their continuous support in bringing this issue successfully.
On behalf of publication committee, I wish to express my sincere gratitude to all authors and reviewers for their contributions and
support to this issue.
I hope this issue will be successful in providing various aspects of Medical Image Processing to IT community. The next issue of
CSI Communications will be on the theme “Bioinformatics”. We invite the contributions from CSI members who are working in
the area of Bioinformatics.
Finally, we look forward to receive the feedback, contribution, criticism, suggestions from our esteemed members and readers at
Wishing a very Happy Teachers day to all Teachers,

Dr. Vipin Tyagi


Dr. Vipin Tyagi, Jaypee University of Engineering and Technology, Guna - MP,
4u u


Message 01 September 2016

Dear CSI members,

From the beginning of this year, we have been trying to increase our reach both
in India and abroad. We are trying to attract professionals to CSI from all parts of
India and trying to motivate existing members to spread a word to their friends
conveying the benefits of CSI Membership.
The different avenues (like IFIP, SEARCC etc.) where our members can participate,
have been opened up. The process of selecting our members to the different
Technical Committees of IFIP from the large number of applications received is
in the final stage. CSI representatives to BIS (Bureau of Indian Standards) have
been assigned work on different ISO drafts by BIS officials. We have informed our
members about ICANN57 to be held in Hyderabad during November 3-9, 2016. ICANN is a very important conference
and the deliberations will be useful for our members involved in network administration. The MOU between IEEE and
CSI for CSI-IEEE CS Education Award is in the final stages and will be announced soon.
I started the month of August, by meeting Ms. Bernie Scott of British Computer Society (BCS) and explored conducting
certification programs of BCS. I was invited to inaugurate an International Conference at Jaipur during August 6-7.
I was invited to deliver a keynote address at NITC 2016 at Colombo during August 8-10, 2016. Glimpses of the
conference have been given separately in this issue. During my visit to Colombo, I had the privilege of meeting and
interacting with Office Bearers of Computer Society of Sri Lanka during the conference and was convinced that there
is enormous scope for cooperation between the two countries. On August 18-19, I was closely involved in organizing
an International Conference on “Internet of Things” at Bangalore which was attended by leading experts including
Prof Raj Buyya (Australia), Prof Sitharama Iyengar (US), Prof Shikharesh Majumdar (Canada), Prof Ernest Cachia
(Malta) and leading practitioners from HP Enterprise, TCS, CDAC, Robert Bosch, Happiest Minds etc. A number of
research papers were presented and the conference served a useful platform for interaction between Industry and
The different Committees formed as per CSI Byelaws have started their work and will be meeting again to take things
forward. We are trying to enlist members in US and Australia. Awards Committee is working on the different awards
of CSI including CSI Fellowship.
Preparations for the Annual Convention of CSI (CSI 2016) to be held in Coimbatore during December 8-10, 2016 is
going on in full swing. The organizers have planned an excellent program with international speakers and leading
industry practitioners. We are inviting all our members to participate in this event.
I am sure with the support and cooperation of all our Members, CSI will continue to play an important role in building
a strong Digital India.
With best wishes.

Dr. Anirban Basu

President, CSI

Dr. Anirban Basu, Bangalore,

u 5 u


Vice President’s Desk
Welcome to the 71st year of Indian Independence!
I feel proud and take this opportunity to remind you that Computer Society of India
(CSI) is serving the nation for more than 50 years in the field of CS&IT out of 70 years
of Independence. Information and Technology (IT) has played a vital role in the nation
development from last three decades. The significance of IT for development was noticed
earlier by CSI and CSI took various initiatives such as IT literacy campaign to rural youth,
disabled and elderly. This has been eventually termed as digital movement from the last few
years. Now a days the governments are keen in promoting this digital literacy as it fetches
or increases the quality of living and other basic necessities of life. CSI step ahead leaves no
stone unturned for promoting the digital mission at the grass-root level.
I hope our esteemed Members of society have experienced the quality of the CSI
Communications July & August, 2016 edition/s. Active steps have been taken with the support
from Esteem Executive Committee of CSI and dedicated staff of CSI to increase the reach
and impact of the journal. CSI is very much active in various social networking sites, I request
all the members and coordinators from region level to institutional level to promote the
activities of CSI through social networking sites and engage more number of students. Persistent improvements are taking place
in the CSI web portal.
We are planning to prepare a roadmap and vision statement, SWOT analysis for the same is under process. Inputs from all
members will be taken into consideration for analysis purpose. The geographical presence of CSI in 19 states and UT’s is less.
Assam, Bihar, Chandigarh, Haryana, Puduchery, Punjab are having poor presence of CSI Membership and Andaman & Nicobar,
Arunachal Pradesh, Dadra & Nagar Haveli, Daman & Diu, Himachal Pradesh, Jammu & Kashmir, Lakshadweep, Manipur, Meghalaya,
Mizoram, Nagaland, Sikkim, Tripura are very poor with regard to membership of CSI. My sincere appeal to all the members to
focus on the CSI membership growth in these states and UT’s.
“A teacher takes a hand, opens a mind, and touches a heart”
CSI is playing a very vital role in Academic Institutions and our esteem Faculties of different Institutions are patronizing
CSI in different capacities since long. On the “Teachers Day”, my regards to esteem Academicians of this country and soliciting
necessary guidance for a vibrant Computer Society of India (CSI).
For feedback & suggestions please write to -
With kind regards

Sanjay Mohapatra
Vice President, CSI

Important Notice
As per the Digital India initiative and directives of the Government of India to Go Green, the Executive Committee of Computer Society
of India in its last meeting held on July 9-10, 2016 at Chennai, has decided to stop the printing of Hard Copy of the CSI Communications,
from January 2017, for all the individual members. The Green India Initiative, which saves both financial and environmental costs and
helps save environment, requires that the CSI Communications be made available to the members through electronic means. This
necessarily requires that members should ensure updating their latest email addresses immediately. Limited number of hard copies
shall be published, for distribution to Authors, Institutional Members and Students’ Branches only, for their Library record. Members,
desirous of still receiving the Hard Copies of CSI Communications, are requested to send their special request, for dispatch of Hard
Copy of CSI Communications, to by October 31, 2016 indicating their CSI membership number.
Such members, who are not receiving the emails of CSI HQ, are also requested to kindly write to and get their email
ID updated, so as to get the CSI publications and other information regularly on their email-id. CSI will not accept any responsibility
for non-receipt of CSI publications or any other information, due to their incorrect email IDs.
Thanking you and looking for your cooperation and support.

Prof. A. K. Nayak
Hony. Secretary
6u u



Medical Image Processing : An Overview

Jyothi Shetty and Niranjan N Chiplunkar
NMAM Institute to Technology, Nitte, Karnatka

Introduction be avoided. Maximum information properties of hair like density, diameter,

Over the years, medical must be extracted. length, level of oiliness etc., These
practitioners have made remarkable ƒƒ Medical image processing algorithms can be used to study hair
contributions to a human’s ability to techniques do not decide, they just properties under various levels of
view inside the body. Medical imaging help to decide. illumination [2].
techniques allow medical practitioners ƒƒ Algorithms must be robust and There is also extensive work going
to take medical images of the parts of safe. Algorithms are expected to on in the field. With image processing
human or animal. Doctors can view the give no answer or a conservative one can define the skin’s health, by
interior parts of the body, which makes answer, rather than a wrong analyzing skin texture image acquired
diagnosis easier. In the last decade, answer. using a UV camera.
there was a tremendous change Image processing may involve Medical image processing
in the imaging techniques, which images like 2D X-ray image of bone, methods are also helping to increase
enabled a transition from qualitative to Complex CT and MRI images of the of IVF implantation success rates [3].
quantitative diagnosis. The significant head or abdomen etc. The use of 3D, The researchers have applied embryo
changes in the imaging techniques 4D images from advanced imaging image processing techniques to implant
have been observed because of advance modalities have made medical image the embryo with the best chance of
in Quantum physics theory (e.g. processing more challenging. Now a success by analyzing the embryos using
NMR), tremendous increase in speed day’s image processing methods even new image segmentation techniques.
and capacity of integrated circuits, support multicore processors and
Different stages of medical image
Revolution in information technology. GPUs [1].
Established methods for clinical The study of medical image
In general, medical image
use are X-ray imaging, Computed processing took off with a lot of hype
processing involves stages such
Tomography (CT), Ultrasound Imaging and now the technology is also maturing
as image acquisition, image pre-
(US), Magnetic Resonance Imaging and good things are emerging from
processing, image segmentation,
(MRI), functional MRI (fMRI), Positron it. Medical image processing is quite
feature extraction and image
emission tomography (PET), Single- helpful in cases where images obtained
photon emission computed tomography from PET, MRI, CT etc, .
(SPECT) etc. Many researchers have applied Image acquisition
Advances in imaging have initiated image processing algorithms on brain Medical image acquisition can be
a lot of technological innovations in images to analyse brain tumor. A brain specified as the method of retrieving
the field of image processing. Medical tumor is nothing but the abnormal growth an image from the different medical
image processing includes the of cells inside the skull. Different image imaging modalities. When the initial
analysis, enhancement and display of processing techniques can be applied CT scanner was invented in 1970’s,
images captured via different imaging on the brain image to find the tumor it took lots of time to acquire one
modalities. Modalities used for medical size, position, etc. Similar liver studies slice of image data and more than 24
imaging produced a list of modern have been done in liver disease. With hours to get a single image out of the
methods for finding the information liver damage too, the tissue undergoes acquired data. Today, this acquisition
about tissue composition, tissue change. This can be studied well using and reconstruction occur in less
detection, analysis of tumor, etc. With medical image processing techniques. than a second. In image processing,
such details, doctors feel diseases In fact, Medical image processing has image acquisition must be always
can be treated at an early stage and simplified a lot of complex issues in accomplished as the first step because,
holistically. Today medical image medical images. Apart from these well- without an image, no processing is
processing has evolved as a field of inter characterized diseases, recent studies possible. The image that is acquired
and multidisciplinary activities, which have increasingly suggested image is completely unprocessed. Enhanced
has become very important in recent processing techniques on DNA, skin image can be obtained by using image
years. Specificities of medical image and hair. processing techniques. Figure 1 shows
processing are, On hair, image processing studies the images obtained from popular
ƒƒ Data are rare and expensive. As far have found that more accurate therapy modalities like MRI, PET and CT
as possible approximations must can be suggested by assessing the scanner.

u 7 u


prostate by segmenting the areas. The results of segmentation
acquire one slice of image data and more than 24 hours to get a single image out of the acquired
and MRI image of the knee is as shown in Figure 3. Thresho
data. Today, this acquisition and reconstruction occur in less than a second. In image processing,
clustering are some common method used in segmentation. Re
image acquisition must be always accomplished as the first step because, without an image, no
COVER STORY effective use of data mining, neural network, fuzzy techniq
processing is possible. The image that is acquired is completely unprocessed. Enhanced image can
images. Fuzzy techniques are very effective in the segmentatio
be obtained by using image processing techniques. Figure 1 shows the images obtained from
regions of these images are not well defined.
popular modalities like MRI, PET and CT scanner.

prostate by segmenting the areas. The results of segmentation

(a) (b) (c) and MRI image of the knee is as shown in Figure 3. Thresho
Fig.1 : a) MRI showing blood vessels in the brain [4] b) PET image [5] c) CT scanner image [5] clustering are some common method used in segmentation. Re
effective use of data mining, neural network, fuzzy techniq
Image pre-processing and CT scans are difficult to study
The pre-processing in image manually, so more effective image images. Fuzzy techniques are very effective in the segmentatio
processing is required to upgrade the segmentation approaches are required. regions of these images are not well defined.
visual appearance of images. Different Segmentation can be mainly used in (i) (i)
types of pre-processing include separating different tissues from each Figure 3: i) Result of watershed segmentation a) original C
image resampling, grayscale contrast other. Various segmentation techniques
c)Smoothed lung Borders d) lung regions [7] ii) a) Origin
enhancement, noise removal and have been proposed by the researchers
mathematical operations. Using Image to detect lung cancer on CT images segmented images using clustering method with different param
resampling number of pixels of the and also to segment lung nodule.
dataset can be reduced or increased. Other applications of segmentation
Grayscale contrast enhancement include simulation of blood and heart Feature extraction
improves the visualization by brightening flow during surgery, finding disorders The main purpose of the feature extraction is to extract the
the dataset. Noise removal being the in several parts like brain, heart, knee,
popular pre-processing step, makes (b) prostate by segmenting
liver, (c)the areas. medical image sometimes input data to a required algorithm
use of several types of filters to remove The results of segmentation applied contain large volume of data, but less information, then the re
noise 1: a)the
from MRIimages.
blood vessels
used inin the on
brainthe[4]CTb) PET
image image [5] lung
of the c) CTand scanner
MRIimage [5]
extracting set of features. This transformation is called featur
the literature to preprocess the medical image of the knee is as shown in
depicted in Figure 2. Medical Fig. 3. Thresholding, region growing, the input file mainly involve shape, intensity and texture featu
image after applying mean filter is also classification, clustering are some be used in applications which help to detect cancerous nodule
shown in Figure. The
The pre-processing mathematical
in image common
processing is required to method
upgrade used in segmentation.
the visual appearance of images/
are also applied on (i)brain
(ii)MRI images to predict the volume of
operations can be used to enhance Research has been directed towards
Different types of pre-processing include image resampling, grayscale contrast enhancement, noise
the particular features. It is possible the effective use of data mining, Figure Fig. 3:3:i) Result
i) Result of watershed
of watershed segmentation a) original C
apply and mathematical
operations Using Image
neural resampling
fuzzy oftechniques
pixels of theindataset a)can
c)Smoothed lung Borders d) lung regionslung
CT image b)Segmented [7] ii) a) Origin
Image registration
addition, subtraction and morphological the segmentation of medical images.
be reduced or increased. Grayscale contrast enhancement improves the visualization by brightening
borders c)Smoothed lung Borders d) lung
operations like dilation, erosion on to Fuzzy techniques are very effective in segmented regions
Image images
[7] ii) a)
registration using clustering
is the methodMRIofmethod
of knee with different
superimposing two orparam
the images.
the dataset. Noise removal being
Mathematical the popularthe
operations pre-processing
segmentation step,of makes use of several types
echocardiographic bone
of b), c) &d) are segmented images
different times, from method
using clustering different positions,
with differentand/or by different m
filters pre-processing
to remove noisecan frombetheused
images. images
to Filter used as in the regions oftothese
the literature images
preprocess the medical
Feature [8]
identify the change in brain size due to are not well defined.
aging andisdementia.
depicted in Figure 2. Medical image after applying mean filter is also shown in Figure. The
Feature extraction butmain
less purpose of the then
information, feature
reducedis to extract the
The mathematical
Image segmentation operations can be used to enhance the particular features. It is possible to apply
representation can be used
The main purpose of the feature medical image sometimes input data to a required algorithm by extracting
arithmetic operationsislikedefined
addition,as a
subtractionextraction is to extract
and morphological thelike
operations required set of features. This transformation
dilation, erosion
contain large feature
is called volume of data, but less
extraction. information, then the re
method, where image is partitioned or portions of an image. In medical image
on to the images. Mathematical operations based pre-processing can be used to identify the change extracted
extracting set from
of the
features. input
This file mainly
transformation is called featur
segmented into several parts based on sometimes input data to a required
the values
in brain ofdue
size pixels. Theand
to aging large images
dementia. involve shape, intensity and
algorithm is very large to be processed. the input file mainly involve shape, intensity and texture featu texture
generated by the modalities like MRI It may contain large volume of data, features. Feature extraction methods
applications which helpwhich
in applications to detect
help cancerous nodule
to detect cancerous nodules in the
are also applied on brain MRI images to predict the volume of
lung. The extraction methods are also
applied on brain MRI images to predict
the volume
Image of tumor in the brain.
Image registration
registration is the method of superimposing two or mo
Image registration is the method
different times, from different positions, and/or by different m
of superimposing two or more images
of the same scene taken at different
(a) (b)
Fig. 2 : Image(a) (b)a) Before b) after applying a mean filter
pre-processing, using Mean Filter [6] times, from different positions, and/
or by different modalities. Image
Figure 2 : Image pre-processing, using Mean Filter [6] a) Before b) after applying a mean filter
8u u

Image segmentation CSI COMMUNICATIONS | September 2016

Segmentation is defined as a method, where image is partitioned or segmented into several parts
registration has been widely used in directly match the features. The information. The days are not far, when
the medical imaging, but that need to feature based registration approaches a mirror at home may give medical
be very accurate. The aim of the image like current and varifold based image warning regarding the changes in our
registration in medical image processing registration are used in the recent face or a person gets an alarm on some
is to integrate the information acquired literature to perform registrations serious medical conditions such as
from different sources to accomplish of MRI and Lung CT images. Image seizures, depending on her/his walking
more complex and detailed data. registration can also help to measure style, which are captured using imaging
how the tumor changes during or after devices. This type of perfect decision
treatment. A long term research goal is possible when advanced image
in the field of image registration is real processing algorithms are in place.
time registration, which has to perform References
required processing in real-time. 1. Anders Eklund, Paul Dufort, Daniel
Researchers have already presented Forsberg, Stephen M. LaConte, Medical
a real time registration procedure, image processing on the GPU-Past,
which helps in mitral valve repair to be present and future, Medical Image Analysis,
performed while the heart is beating. Vol.17, No.8, pp 1073-1094, Dec. 2013.
2. Shih H, A precise automatic system for the
Fig. 4: Row a) MRI with right parietal The recent example also includes use
hair assessment in hair-care diagnosis
glioblastoma, Row b) Corresponding PET of GPU’s for image registration. applications, Skin Res Technol. ;21(4), pp.
layers Row c) Fusion of registered layers [9]
Algorithm specific to individual 500-7, Nov. 2015
modalities and diseases 3. New embryo image processing technology
The rapid development that has
The image processing techniques could assist in IVF implantation, Aug, 2016
taken place in the image acquisition [available at]
devices, made image registration more differ widely based on the application,
0 8 - e m b r y o - i m a g e - t e c h n o lo g y - i v f -
important. Medical image registration imaging modalities and other aspects. implantation.html
can be done on the images taken using Various modalities are used in imaging, 4. Allen T. Newton, Introduction to Modern
a single modality or multi modalities. also on various body parts. MRI and Medical Imaging. PAVE 2014
Single modality based registration CT scanners provide brain, liver, chest, 5. h ttps://
compares the image taken from the abdominal images. Image processing medical-imaging
methods applied to brain tissue has 6. Vincent Luboz, Image Segmentation
same modality on the same patient,
divergent demands from the methods and Pre-processing, [available at]
but may be on different time frames.
But multi modality based registration used in the case of liver. The method
involves images taken from different which gives the correct result with E4F7ED301430/
modalities. The images obtained from respect to one modality may not work 7. Sonali Bhadoria, Preeti Aggarwal, C.
computer tomography (CT) and NMR for another. G. Dethe, and Renu Vig, Comparison
data can be combined using image Conclusion of Segmentation Tools for Multiple
registration techniques to get more Modalities in Medical Imaging, Journal of
In the past, the imaging quality
advances in Information Technology, Vol 3,
information about the patient or monitor was not good and innovations largely No. 4, 2012
tumor growth, treatment verification, focussed on development of new 8. Alireza Norouzi, Mohd Shafry Mohd
etc., Similarly MRI or CT data can be materials. The development of high Rahim, Ayman Altameem, Tanzila Saba,
compared with PET data which has dimensional techniques and high Abdolvahab Ehsani Rad, Amjad Rehman
localized picture of the brain area, to get quality imaging techniques such as high and Mueen Uddin, Medical Image
more valuable additional information for field MRI, CT scanning etc. has made Segmentation Methods, Algorithms, and
diagnosis of the disease. Multi modality new innovations in image processing Applications, IETE Technical Review, Vol.
based registration on MRI and PET data 31, No. 3, pp:199-213, June 2014
more important.
9. Thomas M. Deserno, Biomedical Image
is as shown in Fig. 4. In future, the major goal of the Processing, p 154, ISBN 978-3-642-
Medical image processing often medical practitioner to interpret 15816-2, 2011
involves elastic registration techniques the images better and derive more n
because it has to deal with deformation
of the subject due to respiration and About the Authors:
other related issues. Now a days lung Mrs. Jyothi Shetty is working in the department of Computer Science
CT image registration has become very
& Engineering, NMAM Institute of Technology, Nitte, Karnataka.
important for several observations like
She has published more than 10 papers in reputed journals and
tracing lung motion over the breathing
conferences. Her research interests include cloud computing, Web
cycle, observe anatomical changes
services and image processing. She can be reached at jyothi_shetty@
over time and detecting abnormal
mechanical properties of the lungs.
Medical image registration can be Dr. Niranjan N. Chiplunkar [CSI-00006602] is currently principal
done using intensity-based or feature- of NMAM Institute of Technology. He has completed DST funded
based approaches. Intensity based research project as a principal investigator. He is also associated with
approaches use differences between AICTE-RPS(B) funded research project as co-principal investigator.
intensities between the various images, He can be reached at
whereas feature based approaches

u 9 u



Medical Imaging and Processing

Dushyant Kumar Singh
Asst. Professor, Dept. of Computer Science and Engg., Motilal Nehru National Institute of Technology, Allahabad (UP)

The visual presentation of internal tissues. The delivered energy is the number of heart valves and the
body parts is required when those body absorbed by the biological tissues. The sac of the heart to be seen. The echo-
parts are difficult to be visible from eyes heat formed leads to transient wideband cardiography uses 2D, 3D and Doppler
in normal. In such case, images are ultrasonic emission that generate imaging to create pictures of the heart
captured by some means or instruments ultrasonic waves which are then and shows the blood flowing through
to perform the clinical analysis and detected by the ultrasonic transducers the each valves of the heart. It is widely
medical intervention. Use of images and analyzed to produce images. used in chest pain, shortness of breath,
for clinical analysis and diagnosis is Thermography: It’s an example cancer treatment etc.
termed as medical imaging. Medical of medical infrared imaging. The most Positron Emission Tomography
imaging attempts of uncovering the widely used area of thermography is
(PET): Positron Emission is a nuclear
internal structures of body hidden by like pain diagnostics, breast imaging in
medicine, functional imaging technique
bones and skin. A number of imaging case of breast cancer, and various early
which is used to observe metabolic pro-
techniques has now been evolved which stage disease screening.
phenomenally works well in study of Tomography: Tomography is a cess in the body. The system detects
complex medical diagnosis. Some of medical imaging procedure that uses pair of gamma rays emitted indirectly
those are Gamma-Ray Imaging, X-Ray special X-Ray equipment to create by a positron emitting tracker, which is
Imaging, Ultrasound etc. detailed image or scans of area introduced into the body on biological
In Gamma-Ray imaging, patient inside the body. This technique uses a active molecules. 3D images of tracker
is injected with radioactive isotope tomographic optical system to obtain concentration within the body are then
emitting gamma-rays. With the decay, tomographic images of a specific cross constructed by the computer analysis.
images are produced from the emission section of a scanned object allowing the After a brief note on available
collected by the gamma-ray detectors. user to see inside the object without imaging techniques, the next part
X-rays are electromagnetic radiations making any changes in the actual of discussion follows processing of
and are generally used for bone object. images generated as a result of medical
identification in medical imaging. The Electrocardiography (ECG): ECG imaging. The study or analysis of
Ultrasound imaging uses high frequency is used to record the electrical activity behavior present in images can be done
sound waves (approx in MegaHz) to of heart by using electrode placed on manually by the experts and doctors.
create an image. It is safe and painless human skin. The electrodes by observe
the experts and with
While doctors. While withofthe
the orientation orientation of technolo
and it does not ionizing radiation. The and detect the tiny electrical changes
images for some
towarddiagnosis is done with
automation, the computers.
task of For this, ima
ultrasound scan is an event and the in the human skin that arises from The
diagnosis. analyzing
processingimages for some
of medical diagnosis
images is with an exam
is detailed
sonogram is the image produced when heart muscles. It is a very commonly
done with computers. For this, images
an ultrasound scan is performed. Some performed cardiology test. The graph
Image Processing on MRI Images
are processed for particular objective
other imaging techniques that have of voltage and time is generated by the
of diagnosis. The processing of medical
their promising role in medical domain non-invasive medical imaging process
Here we will images
see how isimage processing
detailed can example/
with an be done to detect tumor
are next discussed. is referred to electrocardiogram.
Tactile Imaging: It is a medical Magnetic Resonance Imaging
from MRI in two
case ways,
of MRIeither darker
images. in color than brain tissue shown
imaging technique that uses touch figure
(MRI): MRI machine emits the radio 2. Image Processing on MRI Images
or sense of touch to convert to digital frequency pulses at the resonant
image. It is a function of P(x, y, z), frequency of the hydrogen atoms in the
where P is the pressure and x, y, z are water molecules. The radio frequency
point coordinates. It is used to map the antenna sends the pulse to the area
pressure or sense of touch on the soft of the body to be examined. The
tissues. Breast, prostate, pelvic and MRI creates a 2D image of thin slice
muscle are application location of this body which is later examined by the
type of imaging. tomographic technique.
Photoacoustic Imaging: is a non- Echocardiography : The ultrasound
invasive medical imaging technique of heart is referred to the echocardio-
which allows structural, molecular and gram. It allows detailed structures of
functional imaging. It delivers non- the heart including the chamber size of
ionizing laser pulses into biological the heart, function of the heart, shows Figure
Fig. 1 1: T1weighted
: T1 weighted scan

Although radiologists have all these MRI images, but due to lack
10 u u
conclusions based on these images vary from doctor to doctor. The
image processing to detect tumor in brain. The basic flow of process i
with the orientation of technology toward automation, the task of analyzing
with computers. For this, images are processed for particular objective of
images is detailed with an example/case of MRI images.

ng can be done to detect tumors in brain. Tumors appear on images obtained
n color than brain tissue shown in figure 1 or lighter than brain tissue shown in

obtain image of brain without affecting contains two stages: preprocessing and
physical qualities of the main image. enhancement of image in first stage
Histogram equalization is used in this and segmentation and classification in
technique for the enhancement of image second stage.
and edge detection and segmentation We can also use machine learning
for getting patterns of brain tumor. In for segmentation of MRI scanned
image segmentation whole image is images of brain in various ways. Most
divided into parts or regions based on commonly used machine learning for
similar features. Another technique these images is a classifier to calculate
hted scan Fig.Figure 2: weighted
2 : T2 T2 weightedscan
scan that can be used for the purpose of weights using training data sets.
segmentation is threshold technique Artificial Neural Networks (ANNs) is
MRI images, but due to lack of any standard process for tumor detection,
in which all adjacent pixels having grey then used for segmentation of new data.
Here we will see how image
ary from doctor to doctor. There are various techniques that can be used in
level, texture, color value etc. lying Neuro fuzzy segmentation process
rain. The basic flow can inbefigure
of process is shown done
3. to detect
between certain ranges belong to same can also be applied for segmentation
tumors in brain. Tumors appear on
class. of images to detect various tissues in
images obtained from MRI in two ways,
Histogram thresholding can also brain like white matter, gray matter,
either darker in color than brain tissue
be used for segmentation which is Cerebrospinal fluid (CSF) and tumor.
shown in figure 1 or lighter than brain
somehow different from one explained This method classifies images layer
tissue shown in Fig. 2.
above. This technique uses the concept by layer and thus high value of tumor
Although radiologists have all
that the backgrounds of images are pixels can be obtained.
these MRI images, but due to lack of any
uniform on which objects are placed If images are colored then
standard process for tumor detection,
irregularly. Thus we can use the traditional techniques cannot fulfill
conclusions based on these images vary
histogram of an image for delineation the segmentation problem completely.
from doctor to doctor. There are various
techniques that can be used in image of object and finding an appropriate Thus for colored images one of
processing to detect tumor in brain. The threshold between object and the latest technology in computer
basic flow of process is shown in figure background for object identification. In science known as soft computing
3. this technique to achieve segmentation is used. Soft computing combinely
input image we can take all pixels having intensity uses fuzzy logic, neural networks and
larger than threshold in one group and genetic algorithm for segmentation
those having smaller in another group. process. Soft computing mainly deals
If we use more than one threshold with approximate models and after
value then technique is called multi processing it gives solution for complex
gure 3: Basic flow of processing MRI image thresholding. problems.
ing tumors is the edge detection technique which can detect boundaries and If we use combined method of
MATLAB based image processing
d also for segmentation. Edges of an segmentation region growing and edge detection for
image are mainly those parts where there of MRI images for tumor detection is
olor present. Hence by using edge based segmentation technique we can obtain the segmentation of image, we can shown below. Following are the steps
ysical qualities of the main image. Histogram equalization is used in reduce this the errors in segmentation performed:
characteristics arise due to separate 1. Input an MRI Image of brain
detection use of region growing and edge 2. Change it grayscale
detection. We can use boundary 3. Apply high pass and median pass
smoothing for ensuring more realistic filters respectively
region boundaries. 4. Compute threshold segmentation
analysis 5. Compute morphological operation
Genetic algorithm can also be
used for classification of brain tissues 6. Get image of tumor as output
in MRI. It uses a special method called Figures numbered from 4 to 9 are
output image spatial gray level dependence in which the image results of the operations
optimal texture features are extracted performed above.
Fig. 3: Basic flow of processing MRI image from normal and tumor regions and Conclusion
show the location of tumor in brain. Imaging adds up to the effective
The mostly used technique in Simple clustering algorithm can also be medical treatments by eliminating extra
detecting tumors is the edge detection used to detect the location and shape surgical operations which are required
technique which can detect boundaries of tumors in the brain. This technique to uncover internal body organs. Imaging
and discontinuities present in an image uses two algorithms for segmentation being non-invasive makes diagnosis
and also for segmentation. Edges of an using computer aided methodology. easier and more flexible. It also prevents
image are mainly those parts where Clustering algorithm such as Fuzzy the many medical risks, especially
there is a rapid change in the grey tone C and intelligent optimization tool when very sensitive/caring organs
of color present. Hence by using edge like genetic algorithm can combine such as brain, back-bone, heart etc
based segmentation technique we can together for this process. This method are considered. Generated images are

u 11 u


Andheri (E), Mumbai-400 093 Maharashtra, INDIA.
Phone : 022-2926 1700 Fax : 022-2830 2133
Email : website :

Application received date : ______________________ Received By : __
Application processed by : ______________________
Membership No.

(Membership subscription fees details for the information of the applicant, not to be attached alo
Application Form to be sent to CSI)
VI. Membership Subscription Fees:
Fig.4 : Input image Fig. 5 : Grayscale image Fig. 6 : HPF image Fig.7 : Median filtered Fig. 8 : threshold segmented Fig. 9 : Extracted
image tumor
1. Individual Membership Fee:
(The membership Period is on Rolling Year basis)
only useful if some automatic analysis for brain tumor segmentation.”  BIME
review.”  Pattern recognition  35.10
of complex medical imagesMembership can be done Category journal 6.1 (2006): One29-33.
Year (2002): 2279-2301.
Two Years Three Years
for diagnosis and treatment. Processing [2]  Ozkan, Mehmed, Benoit
Imaging M.toDawant,
adds up and
the effective [5] 
medical treatments Min, Huang, extra
by eliminating Sun surgical
Bo, and Xi Jianqing.
operations which are required to
Within India Rs.uncover
Robert J. Maciunas. 1000internal
15% Rs. 1800
body organs. Imaging being + 15%
“An makes diagnosis
Optimized Image Rs. 2600
and more + 15%
MethodIt also prevents Rs
these images with different processing
basedService Taxmany= Rs. 1,150.00 Service
the medical risks, especially whenTax Rs. 2,070.00
very= sensitive/caring organs Service
such as brain, Rs. 2,990.00
Tax =Clustering
back-bone, heart etc areService
segmentation of multi-modal Based on Hierarchal
techniques for different ailments helps medical images:
considered. Generated images are only useful if some automatic analysis of complex medical images can be done
a comparative and and with Genetic Algorithm.” 
in effectively diagnosing the Outside
diseases.IndiaA $60
for diagnosis and treatment. Processing these$images
110 different processing $Information
techniques150 for different ailments
(Inclusive of Service Tax) prospective study.” IEEE
helps in transactions
effectively diagnosing on
the diseases. A Technology
very correct diagnosisand
can Applications,
only assert a correct 2009. of disease.
very correct diagnosis can only assert a Medical Imaging 12.3 Example of(1993):
brain tumor534-544. IFITA’09.
is shown in the article as part International
of the processing Forum
of brain images on. Vol.
for tumor 1.
correct treatment of disease. Example
Conclusion [3]  Atkins, M. Stella, and Blair T. IEEE, 2009.
2. Life Membership FeeMackiewich.
(irrespective of “Fully
the age ofautomatic
References the applicant):
Imaging of brain tumor
shown in theby article
adds up to the treatments extra surgical operations which are required to
[6]  Mancas, Matei, Bernard Gosselin,
uncover part
body the processing
organs. of brain makes
Imaging being non-invasive images segmentation of
[1] the brain
diagnosis easier and more flexible. It also prevents in
and V. Rajamani. "A high speed and Benoit
parallel fuzzy Macq. for “Segmentation
c-mean algorithm brain tumor segmentation." BIME
the manyfor tumor detection.
medical risks, especially when very Nationality
sensitive/caring organs such as transactions
brain, back-bone,
6.1 (2006): imaging 
etc are
29-33. 17.1 usingLife Membership
a Fee
considered. Generated images are only useful if some automatic analysis of complex (1998):
medical98-107. thresholding.”  Electronic Imaging 2005.
[2] Ozkan, Mehmed, Benoit M. Dawant, and Robert J. Maciunas. "Neural-network-based segmentation of multi-modal medical
images can be done
for diagnosis Within
and treatment. Processing these images India processing
with different [4]  Egmont-Petersen,
techniques images:
for different a comparative and prospective Rs.
Michael, Dick de
ailments study."10,000 + 15%
IEEE transactions Service
on Medical Tax
Imaging 12.3 = Rs.534-544.
International Society for Optics and
helps in [1]  Murugavalli, S., and V. Rajamani. “A high
effectively diagnosing the diseases. A very correct diagnosis can only assert a correct [3]
Ridder, and medical ofAtkins, M.
Handels. Can
and even
T. be paid
Mackiewich. in
"Fully 4 equal
Photonics, instalments
of the spread
brain overtransactions
in MRI." IEEE 4 years*:-on
Example of brain tumor is shown in the article as part of the processing of brain images for tumor detection. 17.1 (1998): 98-107.
speed parallel fuzzy c-mean algorithm processing with neural networks—a each year Rs. 2500.00 + 15% Service Tax = Rs. 2,875.00 n
[4] Egmont-Petersen, Michael, Dick de Ridder, and Heinz Handels. "Image processing with neural networks—a review." Pattern
References * Note: recognition 35.10 (2002): 2279-2301.
About the Author: i. Three PDCs of the amount Rs. 2,875.00 are to be given in the first year itself, along wit
[5] Min,BIME
[1] Murugavalli, S., and V. Rajamani. "A high speed parallel fuzzy c-mean algorithm for brain tumor segmentation." Huang, Sun Bo, and Xi Jianqing. "An Optimized Image Retrieval Method Based on Hierarchal Clustering and Genetic
Algorithm." Information Technology and Applications, 2009. IFITA'09. International Forum on. Vol. 1. IEEE, 2009.
journal 6.1 (2006): 29-33. Membership Application Form.
Mr. Dushyant Kumar Singh [CSI - I150037] is currentlyshall
[2] Ozkan, Mehmed, Benoit M. Dawant, and Robert J. Maciunas. "Neural-network-based segmentation of multi-modal
ii. Membership working as Assistant Professor, Department
effect, if of
medical Matei, Bernard Gosselin, and Benoit Macq. "Segmentation using a region-growing thresholding." Electronic
[6] Mancas,
be terminated with immediate
Imaging 2005. International Society for Optics and Photonics, 2005. PDCs are not realized.
images: a comparative and prospective study." IEEE transactions on Medical Imaging 12.3 (1993): 534-544.
Science and Engineering at Motilal Nehru
iii. IEEE
[3] Atkins, M. Stella, and Blair T. Mackiewich. "Fully automatic segmentation of the brain in MRI."
Additional Institute
transactions liability,
of Technology Allahabad, Allahabad
on account of any subsequent changes in the (UP). His Tax rule will n
medical imaging 17.1 (1998): 98-107. areas of research interest are Computer Vision,
the member. Image Processing, Embedded Design, Advance architectures
[4] Egmont-Petersen, Michael, Dick de Ridder, and Heinz Handels. "Image processing with neural networks—a review." Pattern
recognition 35.10 (2002): 2279-2301.etc. He can be reached
Outside India at
USD $ 650
(Inclusive of Service Tax)
[5] Min, Huang, Sun Bo, and Xi Jianqing. "An Optimized Image Retrieval Method Based on Hierarchal Clustering and Genetic
Algorithm." Information Technology and Applications, 2009. IFITA'09. International Forum on. Vol. 1. IEEE, 2009.
[6] Mancas, Matei, Bernard Gosselin, and Benoit Macq. "Segmentation using a region-growing thresholding." Electronic
Imaging 2005. International Society for Optics and Photonics, 2005. (ADVERTISING TARIFF)
CSI Student Branch Inauguration at Pydah College of Rates effective from April, 2014

Engineering & Technology, Visakhapatnam Computer Society of India

Unit No. 3, 4th Floor, Samruddhi Venture Park, MIDC, Andheri (E), Mumbai-400 093
Biography Tel. 91-22-2926 1700 • Fax: 91-22-2830 2133

CSI - Communications
Colour Artwork (Soft copy format) or positives are required for colour advertisement
Back Cover Rs. 50,000/-
Inside Covers Rs. 40,000/-
Full Page Rs. 35,000/-
Double Spread Rs. 65,000/-
CSI Student Branch at Pydah College of Engineering & Technology, Centre Spread Rs. 70,000/-
(Additional 10% for bleed advertisement)
Visakhapatnam aas Inaugurated by Dr. S. Pallam Settyt, Professor,
Dept. of CS&SE, Andhra University, Visakhapatnam and Dr. Suresh
Full Page with Bleed 28.6 cms x 22.1 cms
Chandra Satpathy, Chairperson, Division-V, CSI and graced by Dr. Full Page 24.5 cms x 18.5 cms
Ramesh Challagundla, Principal, Pydah College of engineering Double Spread with Bleed 28.6 cms x 43.6 cms
and Technology on 04 August, 2016. Mrs G. M. Padmaja, HOD,CSE Double Spread 24.5 cms x 40 cms
welcomed everybody to PYDAH and she stressed on the need for Special Incentive to any Individual/Organisation for getting sponsorship 15% of the advertisement value
students to associate themselves with organizations like Computer Special Discount for any confirmed advertisement for 6 months 10%
Society of India. Special Discount for any confirmed advertisement for 12 months 15%

All incentive payments will be made by cheque within 30 days of receipt of payment for advertisement.
The inaugural function came to an end by presenting Mementos as a
All advertisements are subject to acceptance by the editorial team.
token of Gratitude to the Chief Guests by Dr. Ramesh Challagundla,
Material in the form of Artwork or Positive should reach latest by 20th of the month for insertion in the following
Principal and a vote of thanks by Mr. M. Kamalakar, Associate month.
Professor, SBC- CSI. HOD’s of different Departments and faculty All bookings should be addressed to :
members of the CSE Department attended the programme. More Executive Secretary
than 250 students were present in the programme. The programme Computer Society of India

was enlightening and inspiring for students. The program was a Unit No. 3, 4th Floor, Samruddhi Venture Park, MIDC, Andheri (E), Mumbai-400 093
Tel. 91-22-2926 1700 • Fax: 91-22-2830 2133 Email:
grand success and paved way for many such endeavors in future.

CSI Communications | June 2016 | 42 w

12 u u



Pathological Brain Detection Systems:

A Review
Deepak Ranjan Nayak, Ratnakar Dash, Banshidhar Majhi and Pankaj K. Sa*
Department of CSE, National Institute of Technology Rourkela

1. Introduction Figure 1. These diseases cause severe of protons per unit tissue. In general,
Medical image processing has problems and sometimes lead to death. T2-weighted images have been utilized
encountered striking development Therefore, it is essential to develop for detecting abnormal brain. The most
and has been a vital field of research an early diagnosis system to arrive at discriminating features of a human
in the last few decades. Experts from correct clinical decisions. This process brain are the symmetry which is clear
various disciplines like mathematics, of detecting diseases in the brain is in the axial and coronal brain magnetic
computer science, statistics, biology, usually dubbed as pathological brain resonance (MR) images. In contrast,
and medicine find motivation in this detection (PBD). Magnetic resonance asymmetry in the axial images strongly
area of research. Various medical imaging (MRI) being a non-invasive designates the abnormality/disease.
imaging modalities such as Computed and faster medical imaging technique Hence symmetry in axial MR images
Tomography (CT), Magnetic Resonance produces rich information about the is an important feature that needs to
Imaging (MRI), Positron Emission soft tissues of the human brain and be considered in deciding whether the
Tomography (PET), Electrocardiography thus has been used in solving many MR image at hand is a normal or an
(ECG), X-ray, Ultrasound etc., have PBD problems. The MR images are of abnormal brain.
been developed where digital image different types, namely, T1-weighted, 2. Pathological Brain Detection
processing algorithms are applied T2- weighted and proton density (PD) Systems (PBDS)
to improve the image quality. These weighted MR images as shown in The enormous volume of
imaging techniques provide the Figure 2. T1-weighted MR images are information about MRI makes the
visual representation of the internal used for visualizing normal anatomy, manual interpretation process
structure of the human body which whereas, T2-weighted is useful for troublesome, costly, and time-
has potential applications in noise visualizing pathology. PD weighted consuming. This motivates the
filtering, 3D visualization, registration, images give information on the density researchers to model computer-aided
segmentation, and classification.
Different modalities have their own pros Categories Disease name
and cons over different organs of the
body and no such generalized modality Cerebrovascular
exists to extract relevant information 1. Chronic subdural hematoma
diseases (stroke)
from all the organs. Therefore, it is
necessary to know the suitable type of
modality before designing an algorithm
for a specific disease. 1. Meningioma
The rate of growth of brain diseases Neoplastic diseases 2. Glioma
across all corners of the earth has been (brain tumor) 3. Sarcoma
exponential in the past decade. As per
the estimation provided by the National
Brain diseases
Brain Tumor Foundation (NBTF) in the
United States, it is found that brain 1. Alzheimer’s disease (AD)
tumors are the cause of one-fourth of Degenerative 2. AD plus visual agnosia
all cancer deaths in children. In 2014, diseases 3. Pick’s disease
World Health Organization (WHO) 4. Huntington’s disease
reported that around 2, 50,000 people
were diagnosed with primary brain
tumors in each year across the world. 1. Cerebral toxoplasmosis
The brain diseases can be classified Inflammatory or
2. Multiple sclerosis
into four types, namely, cerebrovascular Infectious diseases
3. Herpes encephalitis
diseases (stroke), neoplastic diseases
(brain tumor), infectious diseases, and Fig. 1: Types of brain diseases
degenerative diseases as outlined in

u 13 u


Fig. 1: Types of brain diseases

Fig. 2: Types of MRI scans

(a) (a) Before pre-processing
Before pre-processing (b) Afte
The enormous volume of
Fig. 2: Types of MRI scans information about MRI makes the manual interpretation
Fig. 4:the Pre-processing using CLAHE
diagnosis (CAD) systems with the goal costly,
process troublesome, ƒƒ Imageand acquisition
time-consuming. This motivates researchers to
ogical Brain model
of assisting computer-aided
radiologists Systems diagnosis
to take correct (CAD) systems with
ƒƒ ROI segmentation 2.2.the goalsegmentation
ROI of assisting radiologists
and fastertodecisions over unknown MRI
take correct and fasterƒƒ decisions over unknown MRI data. CAD
data. CAD systems have received much Feature extraction Segmentation of systems
an imagehave or defining the
more attention much more
in detecting various attention in detecting various diseases with the assistance of
ƒƒ Feature reduction , and crucial and the most time consuming stage
diseases with modalities.
the assistance In fact,
of different CAD has become one of the major research subjects in
modalities. In fact, CAD has become ƒ ƒ Classification.
medical imaging and diagnostic radiology. Although early divides an image
attempts into multiple parts based
at computerized
one of the major research subjects
in medical analysis
imaging of and medical
2.1 Pre-processing
diagnostic images were made in the 1960s, identifyserious
the desired and objects
systematic and other releva
The quality of acquired MR images (b) After pre-processing
radiology. investigation
Although earlyonattempts CAD began in thein
methods (b)
a fundamental
degrade due
changeAfterin pre-processing
as thethresholding
concept methods
at computerized analysis of medical
in the 1960s,of the computer output,
to improper lightening from
or segmentation
the presence computermethods
Fig. 4: Pre-processing using
diagnosis (k-means,
to fuzzy
images were serious of noise. Fig. 4: Pre-processing
Therefore, it is essential using CLAHE
and systematic investigation on diagnosis
CAD to[1]. These systems
pre-process the imageare
prior to termed region growing), texture metho
as pathological
began in the 1980s with a fundamental 2.2. 2.2 ROI segmentation
brain detection system (PBDS) ROIextraction.
feature segmentation
when itFigure 4 depictslearning
is applied to brain methods
Segmentation ofThis image or neural net
an study
change in the concept for utilization of an original pathological brain and its
the computer output,thefrommotivation
automated and philosophy ofofautomated
Segmentation an imagePBDS
pre-processing result using contrast
etc. withoften
or defining
defining its the
the current
region status
of forofobject
interest and
interest (ROI) i
computer future is very crucial and the most time
diagnosis in clinicallimited
to computer- diagnosis.
crucial andhistogram
adaptive the most time consuming
equalization consumingstage
segmentation of inainimage
stage image
analysis.analysis. This
This brain MRp
aided diagnosis [1]. These systems
The block
are commonly termed as pathological diagram indicating
(CLAHE) one
scheme. It PBDS
can be seen
divides an image into multiple is shown
that inprocess
parts based
observed 3.
that the
divides The basic
on regions
an steps
certain distinctions.
image into which contains
multiple It a
the diseased regions in the right-hand parts based on certain distinctions. It
in a PBDS
brain detection are when it side
system (PBDS) identify
image the desired
are more objects and
distinguishable other
aids relevant
to identify information
the segmentation
desired objects andininstead
an image.
of t
is applied to brain imaging. This study than left-hand side image. Hence, pre-
presents the motivation and philosophy
• Image acquisition methods such as thresholding
processing helps in extracting relevant
methodsto (Otsu’s
it is advisable
other relevant extract features
information in anmethod),from
image. color
Many methods such as thresholding
of automated PBDS with its current segmentation
features from the MR methods
images. Many (k-means,
regions. k-means),
(Otsu’s method), transform m
status and future potential • ROI segmentation
in clinical filtering based techniques like median
diagnosis. (watershed, region growing), texture segmentation methods methods(texture filters), and m
The block diagram•indicating Feature oneextraction
filtering and its variants are also being fuzzy k-means), transform methods
to reducemethods
the noise. (pulse-coupled neural network
(watershed, region growing),(PCNN) and its varia
PBDS is shown in Figure • 3. Feature
The basicreduction , and
steps in a PBDS are etc. are often used for objectmethods segmentation. Figure
(texture filters), 5 shows a res
and machine
• Classification.segmentation of a pathological brain MR image using PCNN. It m

Image acquisition
observed Rio
that the Feature
regions which Feature
contains the suspicious abnormali
& Pre-processing extracted through Extraction
Segmentation segmentationReduction instead of the undesirable regions. The
it is advisable to extract features from these identified regions than

Interpretation and (a) Before Performance

segmentation (b) Afte
Formulation of
Fig. 5: Segmentation using PCNN
2.3. Feature Extraction
Fig. 3: Overall block diagram indicating a PBDS
Fig. 3: Overall block diagram indicating a PBDS Feature can be defined as the individual m
being observed. Feature extraction is a tech
2.1. Pre-processing
u u

C S I C O M M U N I C A T I O N S | S e p t e m b e rinteresting
2016 parts of an image as a comp
The quality of acquired MR images in the reallearning datasetsalgorithms
may degrade due toclassification
2.2. ROI segmentation
Segmentation of an image or defining the region of interest (ROI) is very
crucial and the most time consuming stage in image analysis. This process
divides an image into multiple parts based on certain distinctions. It aids to
identify the desired objects and other relevant information in an image. Many
methods such as thresholding methods (Otsu’s method), color-based
rocessing (b) Aftermethods
pre-processing (k-means, fuzzy k-means), transform methods
(watershed, region growing),neural
learning methods (pulse-coupled texturefilters, wavelets,
methods and colorfilters),
(texture histograms. and cons. However, choosing a correct
and machine
network (PCNN) and its variations), etc. However, the study on pathological classifier directly is still an active area
using CLAHE
are often (pulse-coupled
used for object segmentation. neural
brain network (PCNN)
detection reveals that and its variations),
wavelets of research.
etc. Figure
are 5often
shows aused
result offorsegmentation and its variants
object segmentation. like lifting
Figure wavelet, a result
5 shows of
2.6 Performance evaluation
of a pathological brain MR image using un-decimated wavelet, and dual-tree In biomedical image processing,
mage orsegmentation
PCNN. It may theofbe a observed
that the brain
of interest (ROI) MR
complex image
very haveusing
iswavelet PCNN. It pathological
been commonly may be brain images are treated
observed that which contains
the regions the suspicious used
which contains for feature extraction.
the suspicious abnormality are classes; while healthy
as positive
time consuming
image analysis.
through This process
2.4 Feature reduction brain images are served as negative
multiple parts through
segmentation onsegmentation
instead certain instead It
of the undesirable of the reduction
to regions. Therefore,
stage addresses classes. The performance of a classifier
it isregions. Therefore,
advisable to it is advisable
extract to
features from these
the issue identified
of removing regions
irrelevant and than other by a confusion matrix
is determined
bjects and other
extract relevant
features from information in an image.
these identified Many
redundant features from the data. On which involves four elements, namely,
hresholding methods
regions than other(Otsu’s
regions. method), the color-based
other hand, reduction of features true positive (), false negative (), false
ds (k-means, fuzzy k-means), transform saves methods
memory storage, computation positive (), and true negative (). Several
time, and increase comprehensibility. performance measures like sensitivity,
owing), texture methods (texture filters), and
It canmachine
be categorized into two types, specificity, precision, and accuracy
namely, feature selection and feature
lse-coupled neural network (PCNN) and its variations), have been usually used are defined as
transformation. In feature selection, a follows.
for object segmentation. Figure 5 shows subseta result
of theof features of the original Table 1. Performance measures
athological brain MR image using PCNN. It may beselected; however, in
features are
Measures Definition
feature transformation, the data is
gions which contains the suspicious abnormality
transformed are from the original high Sensitivity TP⁄((TP+FN))
mentation instead of the undesirable regions. Therefore, feature space to a new Specificity TN⁄((TN+FP))
space with reduced dimensionality. Precision TP⁄((TP+FP))
ract features from these identified regionsFeature
than other
selection includes filter
Accuracy ((TP+TN))⁄((TP+FN+FP+TN))
based methods, wrapper methods
(a) Before segmentation Sensitivity calculates the number
and embedded methods. Feature
(a) Before segmentation (b) After segmentation
transformation of pathological samples correctly
techniques like
predicted by the model out of the total
Principal Component Analysis (PCA),
Fig. 5: Segmentation using PCNN number of pathological samples;
Linear Discriminant Analysis (LDA),
however, specificity defines the number
and t-Distributed Stochastic Neighbor
2.3. Feature Extraction of healthy samples correctly predicted
Embedding (t-SNE) are more commonly
Feature can be defined as the individual measurable property of theout
used in PBDS. of the total number of healthy
samples. Accuracy is the total number
2.5 Classification
being observed. Feature extraction is a technique
In this stage awhich efficiently
classifier is used to represents
of samples that are correctly predicted.
interesting parts of an image as amapcompact feature
a reduced set vector.
of features Any machine
to a class 3. Materials
label. This mapping specifies during Data acquisition is the primary
learning algorithms perform classification utilizing
the training phase of the the extracted
classifier. features.
and the most important stage of any
There are common feature extractionWithtechniques
the advent ofavailable in the literature.
complex features, image analysis task. Table 2 lists some
many classifiers have drawn attention publicly available datasets and their
(b) After segmentation
entation (b) After segmentation among researchers, namely, k-Nearest links which are often used in this field of
Fig. 5 : Segmentation using PCNN Neighbors (KNN), Naïve Bayes, Feed research. All these datasets include MR
ntation using PCNN forward Neural Network (FNN), Support images of the human brain; however,
Vector Machine (SVM), and Random some contain other modalities like PET,
2.3 Feature Extraction Forests. Each class has their own pros CT etc.
Feature can be defined as the
d as the individual measurable
individual measurablepropertyproperty of of the process Table 2 : List of some publicly available datasets
the process being observed. Feature
re extraction is a technique
extraction which efficiently
is a technique which represents
Dataset Name URL

an image efficiently
as a compact
parts Any machine
Harvard Medical School Data
of an image as a compact feature
perform classification utilizing
vector. Any machine the extracted
learning algorithms Openfeatures.
Access Series of Imaging
Studies (OASIS)
eature extraction techniques available in the literature.
perform classification utilizing the
extracted features. There are common Alzheimer’s Disease
feature extraction techniques available Neuroimaging Initiative (ADNI)
in the literature. Some of them are
Histogram of Oriented Gradients (HOG),
Speeded Up Robust Features (SURF), Internet Brain Segmentation
Local Binary Patterns (LBP), Gabor Repository (IBSR)

u 15 u



4. Existing PBDS purpose of the research for automated 6. Conclusion

minute abnormality detection without Pathological brain detection
In the past decade, researchers
human error is lost. This paves the path systems (PBDSs) aim at predicting
have developed various PBDSs. Existing
PBDSs could be categorized into two for in to face the certain challenges in abnormalities in the brain correctly
types based on the data dimension. this area of work which needs to be kept with the help of MRI scanning. Because
First one includes three-dimensional in mind for any research advancements. of the vital advantages of PBDS, it
data which require scanning of the ƒƒ The first and foremost challenge has gained much more popularity in
whole brain, and the second category we face is, the features extracted medical society in past decades. PBDS
selects slices (2D) from the 3D data. from the images of the brain are will help shift the focus on cure rather
In 2006, Chaplot et al. [3] have first not translational and rotational than diagnosis results. PBDS will
developed a PBDS on 2D MR images invariant. As a solution, focus provide fast detection of the stage of
where 2D Discrete Wavelet Transform should be made on those feature the disease and suggest the physician
(DWT) has been used for feature extraction techniques which are with the pre-conditions and present
extraction. They used Harvard Medical rotation or translation invariant. symptoms of the patients from where
School Data (HMSD) for their experiment In addition, these techniques must on, further diagnosis can be made more
and applied SVM for classification of provide more directional selectivity. efficient. PBDS has unraveled the new
the unknown samples. Later, many ƒƒ As mentioned above, the size era of diagnostics of human diseases,
methods have been developed based on of a small dataset may give giving intricate insight into human and
2D DWT and different classifiers. Wang us satisfactory result, but its animal brains and may give a solution
et al. [4] in 2015 have proposed a PBDS robustness is questionable. Finding to our evolution theories which are still
based on Stationary Wavelet Transform a standard and large dataset in under debate. Moreover, these systems
(SWT) which is a translation invariant itself a big challenge. Collection can be used to further identify diseases
technique and has potential advantages of scanned photocopies from in other organs of the human body.
than conventional DWT. They have hospitals and private organizations References
validated their scheme on HMSD is not easy and prior negotiations [1] 
El-Dahshan, E.S.A., Mohsen, H.M.,
dataset and achieved better results than and deals need to be taken care of. Revett, K. and Salem, A.B.M., 2014.
DWT using PCA and FNN classifier. In ƒƒ Now even if we have the dataset, Computer-aided diagnosis of human
2015, Zhang et al. [5] utilized 3D DWT we cannot train a classifier brain tumor through MRI: A survey and
to extract features from structural MRI. efficiently if we are not provided a new algorithm.  Expert systems with
Their system detected Alzheimer’s Applications, 41(11), pp.5526-5545.
with a sufficient training dataset.
disease and mild cognitive impairment The training dataset does contain [2] 
Doi, K., 2007. Computer-aided
through the kernel SVM classifier. diagnosis in medical imaging:
the information about positive and
historical review, current status and
5. Observations and inferences negative diagnosis, but rarely the future potential.  Computerized medical
situations of early stages of the imaging and graphics, 31(4), pp.198-211.
Based on the literature study, it has
disease take into consideration. [3] 
Chaplot, S., Patnaik, L.M. and
been observed that numerous PBDS
Hence it is a challenging task in Jagannathan, N.R., 2006. Classification
systems have developed in the last
itself to get correctly classified data of magnetic resonance brain
decade. All the current systems are
along with a large volume of it even images using wavelets as input to
validated on a smaller dataset providing
for early detected scans. support vector machine and neural
significant results which might fail to network.  Biomedical Signal Processing
address the complexity and the majority ƒƒ Apart from these, a paramount
and Control, 1(1), pp.86-92.
of real life scenarios. Researchers have amount time and resources is
Wang, S., Zhang, Y., Dong, Z., Du,
been trying to achieve better results by applied upon getting the algorithm
S., Ji, G., Yan, J., Yang, J., Wang, Q.,
combining different feature extraction robust. If current modalities are Feng, C. and Phillips, P., 2015. Feed‐
techniques with various classifiers. not being sufficient, newer ones forward neural network optimized
For feature extraction, wavelet has need to be devised. by hybridization of PSO and ABC for
received much more attention than ƒƒ Another challenge is how to deal abnormal brain detection. International
other transforms. However, the major with 3D images in large numbers. Journal of Imaging Systems and
Technology, 25(2), pp.153-164.
drawbacks in wavelet are: (i) it is To solve this issue, we may use
deep learning architectures like [5] 
Zhang, Y., Wang, S., Phillips, P.,
translation variant, and (ii) it provides
Dong, Z., Ji, G. and Yang, J., 2015.
limited directional selectivity. In convolutional neural network
Detection of Alzheimer’s disease and
addition, the dataset used in most of (CNN), auto-encoders, and deep mild cognitive impairment based on
the schemes contain, only images of belief network (DBN) for better structural volumetric MR images using
the brain of patients who are either performance. These techniques 3D-DWT and WTA-KSVM trained by
in the middle or the last stage of the take images as the input and PSOTVAC. Biomedical Signal Processing
disease. In such cases deformities from are capable to handle features and Control, 21, pp.58-73.
the normal is much more and the very implicitly. n
16 u u



Overview of Medical Image Segmentation in

MR Human Head Scans
R. Siva Shankar K. Somasundara
Assistant Professor, Department of Computer Applications, Professor, Department of Computer Science and Applications,
National Institute of Technology, Trichy, Tamil Nadu Gandhigram Rural Institute-DU, Gandhigram, Tamil Nadu

Image is a collection of pixels with numerical values to represent visual perception of a physical object,
from which the knowledge of a particular object is understood or analyzed. Providing a depiction of
any object or anatomical structures gives more application possibilities. Medical image processing
involves capturing the images, apply filtering in them, enhance them at requirement, segmentation on
interested region, compression with very low level of loss or lossless etc. The goal of medical image
processing is to enable a physician to make a good understanding of information from the images,
quick decision about the nature of disease that afflicts a human organ. Capturing images, processing,
storing and sharing them is a huge pipeline process, but by a finest team work, this will give a giant
leap for India in Future.

1. Introduction: diseases if found. We concentrate here X-rays and suitable formations applying
1.1 Influence of Images in Medical about MRI from imaging modalities tomography method.
Field and few most used methods in the 1.2.2. M
 agnetic Resonance Imaging
Images are used in almost all areas Segmentation of brain images.
in human life cycle. From historical 1.2 Existing Imaging Modalities Magnetic Resonance Imaging uses
period, images are used to denote the Image of tissues in human body internal energy generated by the protons
paths and borders. Each and every one is obtained by a variety of techniques, found in the human body. Protons in
in the world is influenced by images. depending on the nature of the organ the human body will be excited by an
Images of any individual object are to be imaged. The human body consists external RF energy in the presence of
used for identification and to know the of water, fat, bones and other biological a strong static external magnetic field
characterization and properties of the items. Each and every parts in the human in the range of 2 to 4 Tesla. The human
Object. For medical image processing body reacts to external stimulations like body consists mostly of water molecules
we need information in a format which X-rays, radio frequency (RF) radiation,
(H20) with hydrogen atom. The different
computations can be understood and micro waves, ultrasonic waves. Images
tissues found in the human body
done by the experts. Not only that, the are captured by using external energy
produce various order of energy release
diagnosis part is completely with the sources or from internal energy. Some
and that energy is captured as values for
medical practitioners, surgeons and imaging energy sources are harmful
images by different characteristics. The
neurologists. we can’t capture the when exposed to the organs, such as
characteristic of these images depend
information from all the anatomical Gamma rays. Based on the biological
parts of Human or living things in the characteristics or properties such as on the proton density. Fig.1 shows the
same method or technology. Based like fat, calcium, water can be converted human head scan images obtained
on the anatomical properties and into images when imaging modalities from MRI method. MRI is used to get
the participation of the elements or are employed to obtain the images of high contrast and high resolution three
materials influenced in the anatomical the human organ. Imaging modalities dimensional (3D) images of human
part will decide which modality can be also differ, depending on the nature of organ. MRI technique is also used for
used to capture information from the information needed and anatomical, blood flow measurement. MRI called is
part or tissues of the Human Body or physiological or functional [1], [2]. captured and processed with specific
other living things. Images are used in Some of them are X-Ray which can conditions and situations in the name of
basic research, in biological studies, penetrate into the tissues but scattered functional MRI (fMRI) which is used for
and in medical field like monitoring by bones. Computer Axial Tomography studying neural activities. MRI is widely
the growth, finding abnormalities and which is obtained by N number of used for imaging brain tissues.

u 17 u



portion in Human head scans has the preprocessing methods like filtering
possibilities of having errors. They are and enhancement. Filters like median
also slow and are operator biased [3], filters, binary filters and so are in list of
[4]. There are some semi automated maximum used filters around medical
works [5], [6] are based on finding the field.
tumor in the brain portion and brain 2.2 Clustering and Morphological
structures and portions in human operations:
head scans. But these methods have
2.2.1 Clustering Technique
a need of human interaction between
Clustering concept is used in
Fig. 1.a
the processes of segmenting the brain
images for classifying the same type of
portion. Fully automatic methods like
intensity pixels. We are using clustering
[7], [8] were developed to overcome
because the properties of the brain
problems like human interaction and
portion and head scan images are
reduced time of the process. These
having intensity values within a certain
automatic methods give high accuracy
known limit. The numerical values
and new way of automation too.
are involved in any form clustering
There are many other methods
and morphological operations [9]. In
used for segmentation of brain portions
Clustering there are many works based
based on region of interests like region
in it. Few notable clustering worked so
growing, largest connected component,
far are K-Means Clustering which could
binarization, etc. There are some refined
Fig. 1.b be used for grey scale images. K-Means
methods like watershed algorithm,
clustering will add the pixels into a
snake algorithm are some based on
Fig. 1: Fig.1.a and Fig.1.b gives cluster when the pixel property satisfied
edge based segmentation methods.
T1 and T2 weighted Axial view of the a condition or it will purely avoid the
Mathematical based works are there to pixel. Fuzzy C means [10],[11] could be
Human Head Scan images. support and give some more accuracy used for clustering the complicated
2. Need for Segmentation of Brain in segmentation like using graph theory, tissues and analysis with more number
Medical segmentation can be exponential functions [9]; Figure 2 of clustering options based on the
done by the physicians. But in not all shows two set of original images and weight or ranking system in Brain
the cases the physicians can fix a good their segmented images. set properties portion of Human head scans.
successful rate. In another case, less are basically used directly and indirectly
number of operations segmentation in many image processing methods in 2.2.2 Morphological operations
can be done manually. Each and every current scenario. But each method has Morphological operations [2] are
part of human body has its own method its own merits and demerits too. used for processing images based
to segment and not all the time the on the shape. Basic operations in
physician can risk in lives. There meets morphological operations such as
the requirement between the physicians Dilation and Erosion are used either
and Brain segmentation methods. Brain to separate or join regions by specified
Image segmentation is a unavoidable structuring elements as shown in Fig.3
clinical image processing technique are used in these operations. The other
for image analysis and diagnosis morphological operations are opening,
Fig. 2 : Fig 2.a Fig. : 2.b closing, and top-hat and bottom hat
with visualization and quantitative
assessment of the anatomy or tissues which are derived from the basic dilation
in human body. After an image of a and erosion operations.
human head scan is obtained, the
neurologist try to understand the image 0 1 0
features, such as intensity variations,
edges, structure etc. From the image 1 1 1
features a neurologist is able to identify
the location of the tissues which he is Fig. : 2.c Fig. : 2.d
interested to analyses and diagnose. 0 1 0
For that image analysts are trying to Fig.2.a and 2.b shows original
reduce the unwanted parts in the image T1 weighted coronal image and its Fig.3 : A Structuring Element of size 3X3
and give him exactly the region which is segmented image. Fig 2.c and 2.d
intended to look around. shows Original T2 weighted Axial image Erosion operations are mostly
2.1 Existing methods in Segmentation and its segmented image. performed to disconnect weakly
of Brain: Every segmentation method is connected regions. Erosion removes
Manual segmentation on brain having its own requirements and pixels on the object boundaries and it
18 u u



will be achieved by using structuring [2] A.P.Dhawan , “Medical Image Segmentation of the Brain in MRI”,
element. The structuring element is Analysis”, A John wiley & sons, IEEE Transactions on Medical
placed on all pixel positions in the input New York, 2nd Edition, 2011. Imaging, vol.17, pp.98-107, 1998.
image X and it is compared with the [3] Wolfgang Birkfellner, “Applied [9] Siva Shankar.R., “Skull Stripping
corresponding neighbourhood of pixels, Medical Image Processing”, CRC Methods for T1 and T2 Magnetic
and fits the input image. i.e. Y(i,j) = 1 if Press, Taylor and Francis Group, Resonance Images (MRI) of
SE fits X otherwise 0, repeating for all London, 2011. Human Head Scans using Intensity,
pixel coordinates (i,j).
[4] Isaac N.Bankman, “Handbook Clustering and Resonance
Dilation is used to add pixels to Principles”, Ph.D., Thesis, The
the object boundaries and it is achieved of Medical Imaging, Processing
Gandhigram Rural Instititue –
by using structuring element. The and Analysis”, Academic Press,
London, 2000. Deemed University, 2015.
structuring element is placed at all
pixel positions in the input image X and [10] Kalaiselvi T. and Somasundaram
[5] Dubey R.B., Hanmandlu M., Gupta
it is compared with the corresponding S.K., and Gupta S.K.,”Semi- K., “Knowledge based Self
neighbourhood of pixels, i.e. Y(i, j) = 1 if automatic Segmentation of MRI Initializing FCM Algorithms for Fast
SE hits X otherwise 0, repeating for all Segmentation of Brain Tissues
Brain Tumor”, ICGST-GVIP Journal,
pixel coordinates (i, j). in Magnetic Resonance Images”,
vol.9, Issue 4, pp.33-40, 2009.
International Journal of Computer
3. Evaluation metrics: [6] Kevin Karsch, Qing He and Ye Duan, Applications, vol.90, pp.19-26,
Very basic method used by the “A Fast, Semi-Automatic Brain 2014.
physicians are visual perception Structure Segmentation Algorithm
method, they just see it by bare eyes. For for Magnetic Resonance Imaging”, [11] Somasundaram K. and Ezhilarasan
qualitative analysis and also to produce K., “Edge Detection using Fuzzy
IEEE International Conference on
report or exact statistical reports, we Bioinformatics and Biomedicine, Logic and Thresholding”, National
need similarity indexes like Jaccard or pp.297-302, 2009. Conference on Signal and Image
Dice similarity indexes. Processing, pp.157-160, 2012.
The most used and available for [7] Bouchaib Cherradi, Omar
Bouattane, Mohamed Youssfi and [12] Jaccard.P., “The Distribution of
comparing structural similarities of Flora in Alpine Zone”, New Phytol,
two images are Jaccard coefficient (J) Abdelhadi Raihani, “Fully Automatic
Method for 3D T1-Weighted Brain vol.11, pp.37-50, 1912.
[12] and Dice coefficient (D) [13] are
such measurements of asymmetry Magnetic Resonance Images [13] Dice.L., “Measures of the Amount
information on binary regions A and B. Segmentation”, International of Ecologic Association between
where A and B are two data sets. The Journal of Image Processing, vol.5, Species”, Ecology, vol.26, pp.297-
values will compared with each element Issue 2, pp.220-235, 2011. 302, 1945.
in the set and reported with value [8] Stella Atkins M. and Blair n
between o and 1. Mackiewich T., “Fully Automatic
4. Conclusions:
In the field of Medical Image
Processing, Image Segmentation About the Authors:
leads a crucial role and it needs new Dr. R.Siva Shankar [CSI - I1504021] is currently working as
techniques based on new requirements.
It’s not about only new method or Assistant Professor in the Department of Computer Applications,
algorithm; we can try to do more National Institute of Technology-Trichy, Tamil Nadu, India. His
research on hybrid methods too. Very areas of research are Medical Image Segmentation, Distributed
few basic concepts are just reported computing in real time applications. He can be reached at
here, to have basic awareness of the
field. There are many things to be
achieved. We can find a good pipeline of
capturing, pre-processing, segmenting, Dr. K. Somasundaram is working as Professor in the Department of
compressing and storing in future Computer Science and Applications, Gandhigram Rural Institute-
research. Parallel processing may lead DU, Gandhigram, Tamil Nadu, India. He was a visiting research
a good path towards all the things said
fellow in 1987 in International Centre for Theoretical Physics,
Trieste, Italy. His research areas are Medical Imaging, Image
processing, Multi media for teaching, PC Based Instrumentation
[1] G. J. Romanes, “Cunningham’s
Manual of Practical Anatomy”, and Theoretical and Computational Plasma Physics. He can be
Oxford University Press, 15th reached at
Edition, New York, 2003.

u 19 u



Medical Ultrasound Imaging

A discussion on the ultrasound imaging and image quality enhancement
A. R. Anil N. K. Ragesh
Associate Professor, Dept. of CSE, TataElxsi, Trivandrum
Sree Buddha College of Engineering, Alapuzha, Kerala

1. Introduction
Ultrasonic imaging plays very imaging mode, (2) m-mode, suitable to go until we get a crisp image as in CT
important role in Medical diagnosis for assessing moving objects such as or MRI. Most of these studies focus on a
because of its low-cost, non-invasive blood flow using Doppler analysis, and typical noise type called speckle noise,
and real-time imaging properties. (3) Colour mode, with pseudo colouring which is a typical noise that affects
An ultrasound machine transmits to the b-mode image based on the all coherent imaging systems. This is
high frequency sound wave (in 3 to 20 cell motion detected through Doppler particularly due to the excessive and
MHz range) through human body. The analysis. inseparable reflection and scattering
Ultrasonic transducers convert pulses The advantage of ultrasound of the waves from cell locations which
of electrical signals to ultrasonic waves, images being non-invasive, low-cost, are not in focus. The speckle noise
which travel through the body tissues harmless and painless, also has to pay degrades the fine image details and
and some of them get reflected back a price in the image quality. By nature, limits the contrast, which will make
from boundaries where tissue type ultrasound image contains more noise small lesions difficult to detect.
changes – fluid and Soft tissue, Soft contents – especially speckle noise Removal of noise from ultrasonic
tissue and Bone etc. these reflected – due to the flood of reflected waves image can be addressed in two stages
back signals are received at the from all around the subject body. These – (1) noise removal during image
transducers and converted back as noise components have to be removed acquisition and (2) noise removal after
electrical pulses. The delay between to generate a legible image suitable for image is generated. In first case, the
transmission and reception of these diagnosis. Also the skill of the technician imaging system is equipped with extra
pulses are computed to assess the using the transducer probe is also very processing stages to minimize the error
position information. This analysis will important in generating a good image. induced in the image and in the second
be a complex process through which, the The presence of air between subject case, advanced image processing
tissue changes are computed line by line and probe as well as within the subject techniques are applied on the generated
based on the delay in signal reception. will have very bad impact on the image image to suppress the noise and boost
The process of generating the scan generated. Fig. 1 shows the process of information content.
lines (from actuating the transducer ultrasound imaging.
till the generation of scanline) is Ultrasound Image Acquisition process
called beamforming. These scanlines 2. Understanding and removing To discuss the possibilities of noise
generated are then passed through noise from ultrasound image reduction during image acquisition,
different signal conditioning steps to As we discussed, there are different we need to understand how the image
remove the noise as well as to boost the sources of noise in ultrasound imaging. is formed in the acquisition process.
cell information. The processed scan This starts with the loss of contact or air Fig. 1 shows an overview of this process.
lines are then combined to form the final gap between the body and ultrasound Let us discuss this in some more detail.
cross sectional image through a process transducer probe. The beamforming The first stage in ultrasound image
called scan conversion. The final Image process may introduce noise due to acquisition in beamforming, where, the
generated is called ‘Ultrasonogram’ or the excessive scattering of ultrasonic ultrasound transducers are activated
simply an ultrasound image. The choice waves. The signal processing stage with pulses of ultrasonic wave, and
of ultrasonic frequency affects the also introduce data loss while trying then listens to the reflected waves to
quality and depth of the scan. Higher to boost some desired information. generate the scan line. This is repeated
frequencies provide better image There could be noise introduced in for each scan line by adjusting the
resolution but limits the depth to which the scan conversion stage also due to signal strength of the transducers.
the image can be captured, whereas, interpolation. Generation of a scan line is happening
lower frequencies allow to penetrate Analysis of noise in ultrasound in different steps, where in each steps,
to more depth but at a lower image images as well as the techniques to signal pulses are transmitted focussing
resolution. remove them have been subject of at each discrete position (let us call it
There are typically three modes many studies. This study started from as a pixel in final image context) this
of ultrasound imaging – (1) b-mode, the inception of ultrasonic imaging and focussing is done by adjusting the
the basic two-dimensional intensity still continues as there is still long way delay in activating different transducer
20 u u


subject will have very bad impact on the image generated. Figure 1 shows the process of
ultrasound imaging.
the image. For example, changing
the image brightness or contrast or
altering the shape through geometric
transformations. There could be a
variety of basic operations as well as
their combinations to generate different
effects like quality enhancement, object
detection etc. Here in our case, we need
to improve the quality of the image by
removing the speckle noise present in
The speckle noise is typically a
type of multiplicative noise, which are
generally addressed by 2-dimensional
image filters like Median filter, Lee
filter etc. However, the process is not as
simple as applying a single filter. When
the noise is removed, a considerable
amount of the information content is
also will be lost. To avoid this, we need to
find an efficient combination of different
Figure 1: Ultrasound Imaging System Overview linear and non-linear image processing
Fig. 1 : Ultrasound Imaging System Overview
operations, so as to supress the noise
while retaining the image information
2. elements
Understanding in the
removing noise array.
imageQuality Enhancement During
delay is computed based on the distance Acquisition content.
we discussed,
transducer there are to
array different
the focus sources
pointof noise in Theultrasound
qualityimaging. This starts
of generated withcan
image the Active research is happening in
loss of contact or air gap
and the velocity of sound. between the body and ultrasound transducer probe.
be improved during the acquisition by The beamforming this area considering the advantages
processWhen may introduce noise duethe
passing through to human
the excessive scattering
using (1) high of ultrasonic waves. The signal
quality transducers, (2) of the ultrasound imaging techniques
processing stage also
body, a portion introduce
of the data loss
ultrasound while trying
waves to boostthe
ensuring someproper
desired contact
information. There
without over other medical imaging modalities.
get be noise introduced
absorbed by the body in the cells,
scan conversion
which stage anyalso airduebubble
to interpolation.
using gel medium Different solutions are proposed,
will get
Analysis accumulated
of noise in ultrasound andimagesthe waves as well as between
the techniquesthe transducer
to remove them and thehavebody,
been producing variable results mostly
subject of manydeeper willThis
studies. become weakerfrom (3)
study started the appropriate axial and imaging
inception of ultrasonic lateral and
still context sensitive – or in other words,
and weaker.
continues This
as there signal
is still long wayattenuation
to go untiliswe getwhile
a crispdoing
imagethe as insignal processing
CT or MRI. and
Most of these suitable for a particular imaging
studies focus on a by applying
typical noise type a distance
called speckle (4)noise,adaptive
which is interpolation
a typical noise thattechniques
affects all scenario or requirement. A generic
(time) varying
coherent imaging gain value
systems. to isthe
This reflecteddue towhile
particularly doing the
the excessive scan conversion.
and inseparable reflection andA quality enhancement solution, to
signal. This
scattering of theprocess
waves from is called Time Gain
cell locations which aredetailed
not indiscussion of thesenoise
focus. The speckle techniques
degrades produce ultrasound images of quality
fine image details (TGC).
and limits the contrast, whichare will out
make ofsmall
scope of this
lesions article.
difficult to detect. similar to CT or MRI, is far from reality.
Finally, thefrom
intensity at a particular This keeps ultrasound image de-
Removal of noise ultrasonic Image Quality
image can be addressed in twoImprovement
stages – (1) noiseAfterremoval
pixel image
is the delayed sum noising an open and attractive area
during acquisition andof (2)
reflected Acquisition
removal after image is generated. In first case, the
soundsystemwavedisproduced for new research scholars. We are not
imaging equipped by with allextra
the active
processing stages Once the image
to minimize is generated
the error induced in by the
transducers. The summed signal is then discussing these techniques in detail
the scanning system, the quality can
passed to the Signal processing block, here as that is outside the scope of this
be further enhanced by computer
where the signal envelope is extracted, article.
processing using advanced digital
decimated and then log compressed to image processing techniques. This is 4. Performance evaluation of
fit the dynamic range to that required addressed in the next section. different Ultrasound image de-
by the system to generate the image. noising techniques.
Different axial and lateral filters are 3. Digital Image Processing
The performance of different image
applied on the signals at different stage for Quality Enhancement of
quality enhancement techniques is
to improve the image quality. Ultrasound images
compared using different image quality
The processed signal will then be Digital Image processing is
metrics. There are subjective image
given to the Scan conversion module as the study and application of various
quality metrics as well as objective
scan lines, which will actually generate 2- dimensional signal processing
methods. The objective metrics
the image from the scan lines by doing techniques to modify the properties
provides statistical information which
a geometric mapping as shown in Fig. of the digital image. A digital Image is
are good indicators of signal quality,
1. The final image is generated by nothing but a 2-dimensional signal,
however, these metrics treat images as
interpolating the image pixels from the which can be visualized as spatial
signals and does not include the human
mapped scan lines. distribution of luminous intensity and
factor into it. Human perception of
Once a complete scanline is colour information. Different image
image is not just its statistical property.
processed, the same is repeated for processing techniques are used
Methods considering how human brain
next scan line and then the complete depending on what type of alteration
is interpreting the image are more
image. or transformation is required on
suitable in cases like this.

u 21 u



Objective methods are easy Conclusion researchers as attract them to share

to implement and have lesser Ultrasound imaging and enhancing their contributions to this demanding
computational complexity. However, the quality of the generated image are area.
subjective image quality metrics in focus of different research scholars References
without human involvement is for a long time and still there are [1] Ragesh. N. K, Anil A.R, Rajesh R.:
difficult to implement and requires open challenges where more efficient “Digital image denoising in medical
implementation of Artificial intelligence algorithms and techniques are yet to be ultrasound images: a survey”,
techniques, which is computationally developed. Our intention is to bring this ICGST AIML-11 Conference, Dubai,
complex and still evolving. interesting area to the focus of budding UAE, 2011, pp 12-14   n

About the Authors:

Mr. A. R. Anil [CSI-1124860] is currently working Mr. N. K. Ragesh is currently working at TataElxsi. He
as Associate Professor, Department of CSE at is also doing PhD in Computer science at Bharathiar
Sree Buddha College of Engineering, Alapuzha, University, Coimbatore. His research interests
Kerala. He is also doing PhD in Computer Science include Image Processing and Visualization.
at Bharathiar University, Coimbatore. At present he
is serving CSI as CSI Student Coordinator,Kerala
state. He can be reached at

Appeal to all CSI Members

All members of CSI are requested to update their personal details such as mobile number, latest email address, address
for communication and other details in the CSI membership database, if there is any change. This will help CSI to serve its
members better. The change request must be supported by valid supporting proof for the change requested.
The members must provide the following details along with the request:
1. Member’s Name necessary correction / change in details at : hq@csi-india.
2. Membership No. org
3. Old Communication Address with registered email-id The following documents would be accepted for change
(with CSI) and Mobile no. request:
4. New Communication Address with email-id and Mobile Voter ID Card / Aadhaar Card / Passport / Bank (Nationalised)
no. Pass Book with photo / Credit Card with Photo / Driving
Please send the request with any one of the following Licence
document/s duly signed by the member for updating
database at CSI HQ either by registered post at CSI HQ OR Prof​.​ A​.​ K​.​ Nayak
through email to CSI HQ with copy to concerned RVP for Hony. Secretary

Kind Attention: Prospective Contributors of CSI Communications

Please note that Cover Themes for forthcoming issues are planned as follows:
• October 2016 - Bioinformatics • November 2016 - Big Data
• December 2016 - Remote Sensing and GIS • January 2017 - Applications of IT
• February 2017 - Operating Systems • March 2017 - Software Engineering
Articles may be submitted in the categories such as: Cover Story, Research Front, Technical Trends and Article. Please send your contributions
before 20th September for October issue. The articles may be long (2500-3000 words maximum) or short (1000-1500 words) and authored in
as original text. Plagiarism is strictly prohibited.
Please note that CSI Communications is a magazine for members at large and not a research journal for publishing full-fl edged research
Therefore, we expect articles written at the level of general audience of varied member categories. Equations and mathematical expressions
within articles are not recommended and, if absolutely necessary, should be minimum. Include a brief biography of four to six lines, indicating
CSI Membership no., for each author with high resolution author photograph.
Please send your article in MS-Word and/or PDF format to Dr. Vipin Tyagi, Editor, via email id : with a copy to csic@
(Issued on the behalf of Editorial Board CSI Communications)
Prof. A. K. Saini
Chair - Publications Committee
22 u u



Medical Image Retrieval using DICOM Format

Amol Prakash Bhagat Mohammad Atique
Assistant Professor, Innovation and Entrepreneurship Development Centre, Professor, PG Department of Computer Science,
Prof Ram Meghe College of Engineering and Management, Badnera, Amravati. Sant Gadge Baba Amravati University, Amravati.

Medical Image Retrieval using DICOM environment using TCP/IP industry can be unambiguously obtained.
Format standard networking protocol. The interoperability of medical
In 1970’s, variety of digital While the ACR-NEMA Standard imaging devices can be enabled by
diagnostic imaging modalities has was applicable only to a P2P specifying the DICOM Standards:
been invented with the emergence environment. For network based 1. A collection of protocols must be
of computed tomography (CT). With operations it requires a Network followed by devices requesting
the emergence of medical image Interface Unit (NIU). conformance to the standards for
modalities increased uses of the 2. Choice of logical file system of communications in network.
computers in clinical applications are physical media or a file format 2. The commands and related
observed. The requirement of standard to be used for storage and information’s semantic and syntax
method for transmission of images and representation were not specified which can be transferred using
allied information among the devices by the ACR-NEMA Standard. PC these protocols must be specified.
manufactured by numerous vendors file system such as NTFS, FAT32, 3. A collection of media storage
has been identified by American College FAT16, logical file systems ISO services to be utilized by devices
of Radiology (ACR) and the National 9960, industry standard media requesting conformance to the
Electrical Manufacturers Association CD-R can be utilized to support standard, file format, a medical
(NEMA). Different types of digital offline operations in DICOM directory structure utilized to
medical images in variety of formats are standard. simplify access to the images
produced by these devices. 3. The transmission of data is the and related information stored on
ACR and the NEMA formed the joint only objective of the ACR-NEMA exchange media must be specified.
committee. This committee specified Standard. While DICOM standard
standards in 1983: generates the semantics of The following are not specified by
1. Independent of device manufacturer commands and linked data using the DICOM Standard:
encourages transmission of digital the terminologies of service 1. The details of implementing any
image information. classes. features of the Standard on a device
4. A least level of conformance requesting conformance.
2. Enable the implementation and
is provided by the ACR-NEMA 2. A system implemented by
extension of picture archiving and
Standard. The provision is given to incorporating a collection of
communication systems (PACS)
the implementer for structuring devices every requesting DICOM
which can also interconnected
the conformance statement for conformance with overall collection
with other components of hospital
selecting required options in of features and functions to be
information systems.
DICOM standard. expected from.
3. Permit the construction of 5. DICOM standards are provided in a 3. For assessing the implementation’s
diagnostic information databases multi-part document. This enables conformance to the standard with a
that can be accessed by extensive advancement of the Standard in testing/validation.
types of devices distributed a quickly changing environment DICOM Communication Model and File
geographically. by easing the inclusion of novel Description
Different standards are suggested features. ISO instructions for The DICOM Standard belongs to
by ACR-NEMA. These standards are structuring multi-part documents the domain of Medical Informatics. In
maintained by years. These standards have been utilized in the creation this domain the transmission of digital
identified a minimum collection of of the DICOM Standard document. information among medical imaging
software commands, a stable set of 6. Explicit Information Objects are devices and other systems is addressed
data formats, and a hardware interface. introduced in DICOM standards for in DICOM standards. Most of the medical
The numerous key improvements to graphics and images as well as for equipment is interoperable with other
earlier versions of the ACR-NEMA printing, reports, waveforms, etc. medical devices; the DICOM standards
standard are suggested by the DICOM 7. Any Information Object can be should cover all the areas of medical
(Digital Imaging and Communications uniquely discovered using specified informatics. The devices requesting
in Medicine) standard. established technique. The conformance must be interoperable as
1. DICOM standard is applicable to association among the Information per DICOM. Specifically, it:
perform operations in a networked Objects transferred over network 1. Specifies the semantics of

u 23 u


3. Defines the conformance requirements of implementations explicitly for the standard.
Sufficient information must be specified by conformance statement for determining the
functionalities for which interoperability can be estimated with another device requesting
COVER STORY conformance.
4. Enables task in a networked environment and it is organized to accommodate the novel
services, thus enabling support for emerging medical applications.
commands and associated data
for devices to communicate. As per Medical Information Application
standards it specifies how devices Application Entity
react to commands and associated Service Class Specifications
data, not only the information which
is to be exchanged among devices. Information Objects Definitions
2. Specifies the semantics required
for offline communications using Data Set Structure and Encoding –Data Dictionary

the information directories, file

formats and file services. Message Exchange File Format

3. Defines the conformance

requirements of implementations DICOM Upper Layer Service Boundary DICOM Basic File Service Boundary

explicitly for the standard. Sufficient DICOM Upper Security Layer

information must be specified Layer (Optional)
by conformance statement for
determining the functionalities Security Layer
for which interoperability can be (Optional)
estimated with another device
Physical Media
requesting conformance. TCP/IP
and Media File
Transport Layer
4. Enables task in a networked Formats

environment and it is organized to

Network Exchange On-Line Communication Media Storage Interchange Off-Line Communication
accommodate the novel services,
thus enabling support for emerging Fig. 1 : Generic communication model.
Figure 13: Generic communication model.
medical applications. in Fig. 2. As it can be observed from followed after this. For example, the
The PACS solutions can be the Fig. 3 for the DICOM format header file meta information group contains
implemented using the DICOM standard the first 794 bytes are used. These the group 0002hex (on the left in the
but it may not guarantee all the bytes describe image dimensions and example) containing three components:
objectives will be achieved by utilizing maintain text information about the the length of group is defined by first,
DICOM standards only. The generic DICOM scan. file version is denoted by second, and
communication model for the standard the transfer syntax is stored by third.
which extents both media storage offline The type of image decides the
communication and networks on-line required DICOM elements as specified
communication is shown in Fig.1. The in DICOM standard Part 3. For example,
following boundaries may cover all the the group:element 0008:0060, the
applications: modality shown is ‘MR’. The MRI echo
1. The independence from protocols time must be described by its element.
such as TCP/IP and networking The DICOM standards will be violated
communication physical support if this information is not present. The
is provided by the Upper Layer presence of many of these elements is
Service. not checked in reality by various DICOM
2. Independent access to storage viewers such as ezDICOM, MRIcro, only
media for accessing files structures the header information is extracted
and storage formats using the Figure
Fig.14: DICOM
2 : DICOM image
image file. file. by them for knowing the required
Basic DICOM File Service. parameters. The 128 byte header
A header The and type
imageofdata are decides the required DICOM elementsinformation
image as specified and in theDICOM
letters ‘DICM’
Depending on the total information
contained in a single DICOM file. A are not present in the NEMA format.
to be stored the size of header varies.
comprisesPart information
3. For such as
example, the group:element 0008:0060, the modality
Multi-frame shown images is 'MR'.
such asThe 3D images
In this case, an image with 109x91x2
image dimensions, type of image scan, cannot be defined by NEMA.
voxels dimensions is defined in the
patient’s echo time
information, mustID,bescan
patient described by its element. The DICOM standardsThe
header, using 1 byte per voxel data
will “Transfer
be violatedSyntax if this Unique
date, etc. The image data may contain Identification” is defined by the
information is not present. resolution resulting in the total 19838
information in three dimensions. In theThe presence of many of these elements is not checked in reality by
important element “group: element
as image size. After the header image
traditional Analyze format, image data 0002:0010” as presented in table
is maintained one file viewers such asdata
(*.img) and
is present.
ezDICOM, The single
MRIcro, only the file header
is information is extracted by 1. The
utilized for storing the image data and structure of the image data such as
the header data in another file (*.hdr)
them for knowing the requiredtheparameters.
unlike DICOM format. Compression or
The 128 Abyte
header information. 128 header
byte information and the letters
whether it is compressed is reported by
preamble, in which all bits are generally this value. Only uncompressed raw data
‘DICM’ canare
be not
applied to DICOM
present in the NEMA format. Multi-frame images
set to zero, is required by DICOM, the such as 3Dbyimages
is handled most of the cannot
DICOM beviewers.
image data for reducing the size of Both schemes lossy JPEG and lossless
letters ‘D’, ‘I’, ‘C’, ‘M’ are available after
image but byit NEMA.
is not possible with JPEG can be used to compress DICOM
preamble as shown in Fig. 3. Header
Analyze image format. An example of images. In lossy JPEG compression
information organized in groups is
imaginary DICOM image file is shown
The “Transfer Syntax Unique Identification” is defined by the important element some high frequency information is lost.
24 1. The structure of the image data such as
“group: element 0002:0010” as presented in table u u


whether it is compressed is reported by this value. Only uncompressed raw data is handled by

Lossless JPEG scheme is mostly used Table 1: Description of Transfer Syntax bits per sample (65,535 levels), or even
in medical imaging which is basically Unique Identifier. in 12-bit or 32-bit resolution. 16 million
Huffman lossless JPEG not efficient and Transfer Syntax UID Definition colors are described by a RGB image
recent JPEG-LS approach. that stores 3 samples per pixel at 8-bits.
1.2.840.10008.1.2 Raw data, Implicit
VR, Little Endian References
1. A. P. Bhagat and Mohammad Atique, “Medical
1.2.840.10008.1.2.x Raw data, Explicit VR images: Formats, Compression Techniques
x = 1: Little Endian and DICOM Image Retrieval A Survey”, in
x = 2: Big Endian proceedings of the International Conference
on Devices Circuits and Systems, IEEE Explore,
1.2.840.10008.1.2.4.xx JPEG compression pp. 172-176, Mar 2012.
xx = 50-64: Lossy 2. H. Greenspan and A. T. Pinhas, “Medical
JPEG Image categorization and retrieval for PACS
xx = 65-70: Lossless using the GMM-KL framework,” IEEE Trans.
JPEG Information Tech in Biomedicine, Vol. 11, No.
2, March 2007.
1.2.840.10008.1.2.5 Lossless Run Length 3. Pelski, 2009] Sue Pelski, “Oracle Multimedia
Encoding DICOM Developer’s Guide 11g Release 1
(11.1),” May 2009.
The image can be also described by 4. Melliyal Annamalai, Dongbai Guo, Susan
Photometric Interpretation (0028:0004), Mavris, Jim Steiner, “Oracle Database 11g
DICOM Medical Image Support,” An Oracle
the Bits Allocated (0028:0100), the White Paper, February 2009.
Samples per Pixel (0028:0002) in 5. A. P. Bhagat and Mohammad Atique, “DICOM
addition to the Transfer Syntax UID. Image Retrieval Using Novel Geometric
Moments and Image Segmentation
Figure3 :15:
DICOM header.
header. The photometric interpretation is a
Algorithm”, published in International Journal
continuous monochrome, basically of Advanced Computer Research, Vol.3, No. 12,
Table 1: Description of Transfer Syntax Unique Identifier.
The codes presented in table 1 represented with pixels in grayscale, pp. 37-46, Sep. 2013.
Transfer Syntax UID Definition 6. A. P. Bhagat and Mohammad Atique,
are described in the DICOM standards for many of the CT and MRI images.
1.2.840.10008.1.2 Raw data, Implicit VR, Little Endian “DICOM Image Retrieval Using Geometrics
part 5. Transfer
1.2.840.10008.1.2.x Syntax
Raw data, Explicit VR x UID
= 1: Littlereports
‘MONOCHROME1’ with low values
Endian x = 2: Big Endian Moments and Fuzzy Connectedness Image
1.2.840.10008.1.2.4.xx compression and the raw representing bright, and high values Segmentation Algorithm”, published in
data byte order. “BigLossy
xx = 50-64: endian”
JPEG and “little representing dim or ‘MONOCHROME2’ Springer AISC (Advances in Intelligent and Soft
Computing), Vol. 248, No. 2, pp. 109-116, Dec
endian” techniques
xx = 65-70:are used
Lossless JPEG by different with low values representing dark, and
computers for
storing integer values in
Lossless Run Length Encoding high values representing bright are 7. Amol Bhagat and Mohammad Atique,
two different ways. For example, the called as a photometric interpretation “Medical image mining using fuzzy
for describing connectedness image segmentation:
value 257
age can be also described byas a 16-bit Interpretation
Photometric integer: the (0028:0004),
value the Bits Allocated monochrome DICOM Efficient retrieval of patient’s stored images,”
01 is stored by the most significant images. Various photometric in Book Biomedical Image Analysis and Mining
100), the Samples per Pixel (0028:0002) in addition to the Transferinterpretations Syntax UID. The such as RGB, YBR,
byte, while the value 02 is stored by Techniques for Improved Health Outcomes,
etric interpretation theis least
a continuous monochrome,
significant basically represented
byte. Depending CMYK, Palette,
with pixels in etc. are used to Publisher IGI Global USA, Nov 2015.
8. Amol P. Bhagat and Mohammad Atique,
le, for many ofonthetheCTmemory and MRIassignment schemes
images. 'MONOCHROME1' represent color ultrasound images
with low values “Medical Image Retrieval using Fuzzy
some computers save this value as or color medical photographs. Three Connectedness Image Segmentation and
nting bright, and high values representing dim or 'MONOCHROME2' with low values
02:01, while others will store it as 01:02. samples per pixel are stored by color Geometric Moments,” International Journal of
nting dark, and high Signal Processing Systems, Vol. 4, No. 1, pp.
Forvalues representing
matching bright areused
the ordering calledbyasthe a photometricimages interpretation
such as in RGB one each for red,
73-78, Mar 2016.
cribing monochrome computer,
images. viewer
Various may need
photometric green
interpretations such andas blue,
RGB, while only one sample 9. Amol Bhagat and Mohammad Atique, “Novel
to exchange the byte order of the data per image is stored in monochrome Localized Entropy Based Medical Image
CMYK, Palette, etc. are used to represent color ultrasound images or color medical
if data more than 8-bits per sample is or palette images. Most of the images Retrieval,” IETE Journal of Research (http://
aphs. Three samples per pixel are stored by color images such as inare RGBstored
one each
in for
8-bits (256 levels) or 16- 2016.1211494),
present. pp. 1-12, Aug 2016. n
en and blue, while only one sample per image is stored in monochrome or palette images.
the images are stored the(256
in 8-bits Authors:
levels) or 16-bits per sample (65,535 levels), or even
Dr. Amol Prakash Bhagat is presently working as an Assistant Professor in the Department of Computer
Science and Engineering, Prof Ram Meghe College of Engineering & Management, Badnera since December
2010. He is also working on some research projects in Innovation and Entrepreneurship Development Centre.
His research interests include Image Retrieval, Image Processing and Enhancement, Soft Computing, Virtual
Reality, Surface Computing, Network Security and Distributed Computing. He is the author of more than 50
research articles, 3 book chapters and published one patent. He can be reached at

Dr. Mohammad Atique [CSI- 0016254], is presently working as Professor in PG Department of Computer
Science and Engineering in Sant Gadge Baba Amravati University, Amravati since October 2011. His area of
interest includes Image Retrieval, Image Processing, Neural Networks and Soft Computing. He is the author
of more than 70 research articles and published two patents. He can be reached at mohd.atique@gmailcom.

u 25 u



Trend in Medical imaging

Nityesh Bhatt Abhishek Agrawal
Professor, Information Management, Institute of Management, Gandhi Medical College and Hamidia Hospital, Bhopal, M. P,
Nirma University, Ahmedabad,

Rising aspirations, growing stress, Imaging Market on innovation side too, companies are
changing lifestyle and fast food culture At present, the global diagnostic showing their forte.
have led to unhealthy society across the imaging market is dominated by 4. Imaging Market in India
globe requiring medical intervention North America as a result of growing Diagnostic imaging device sector
from very early stage. On other side; in incidences of chronic diseases and large in India is estimated to register robust
last few decades, inventions in medical chunk of aged population. According to growth rate through 2022, due to low
science have grown phenomenally. One a study by Yole Development, the global market penetration and rising medical
such invention is in form of medical market for medical imaging sensors procedures along with government’s
imaging technology that helps medical is projected to grow to 6.7 million support. Major suppliers of high-end
professionals to see inside the body units in 2019 from 2.1 million units in systems in India are GE Healthcare,
and also allows them to do keyhole 2013, exhibiting a CAGR of 11 percent Toshiba Medical Systems, Siemens
surgeries without opening much of the reaching US$ 142 million by 2019. Asia- Healthcare and Philips Healthcare.
body. Newer techniques like CT Scan, Pacific region is projected to grow at IRIA (Indian Radiological and Imaging
Ultrasound and MRI have overtook the a rate of 6% through 2022, with China
Association) is also promoting the
initial technology of X-Ray imaging by and India witnessing fastest growth
research on various imaging modalities
allowing doctors to visualize the interior because of increasing awareness and
using various platforms. A recent study
of the body in three dimensions. Under healthcare expenditure, high patient
by Decision Resources Group’s (DRG)
this backdrop, this paper provides an ratio, and government’s focus on
on imaging diagnostic market in India
overview of various medical image health. Key players in this market are
suggests three key ingredients for the
processing systems and their benefits. GE Healthcare, Siemens Healthcare,
growth of this segment. These are high
Subsequently, it covers global and Indian Philips, Samsung Electronics, Hitachi,
growth of medical tourism due to a low
market of medical image processing Toshiba Corporation, Carestream
cost; upward trend in acquiring high-
systems. Health.
end and advanced imaging systems;
The BRIC (Brazil, Russia, India
1. Medical Image Processing and increasing demand for refurbished
and China) medical imaging device
Systems devices in order to check on prices.
sector is expected to grow by CAGR of
Various image processing systems As per a trend analysis, large MNCs
10.5 percent reaching US$7.6 billion
used today are: are now looking to modify their products
in 2018. This growth rate is attributed
a) Endoscopy to match the requirements of Indian
to the launch of cost-effective imaging
b) Radiography healthcare facilities. For example, GE
solutions, increasing prevalence
c) Contrast Agents Healthcare launched Discovery IQ, a
chronic disorders, and enhanced
d) Nuclear Medicine PET (Positron emission tomography)/
private and public funding in healthcare
e) Ultrasound CT molecular imaging system designed
infrastructure. A 2013 report of Markets
f) Computed Tomography in India is claimed to be 40% more
and Markets says that analytical
g) MRI affordable. Toshiba Medical Systems
software for diagnostic imaging
h) PACS & DICOM. has introduced Vantage Elan MRI system
devices market is forecast to reach
with optional power-saving mode and
approximately US$ 2.4 billion by 2017
2. Advantages of Digital Processing compact design. Entry of Dabur into
from US$ 1.7 billion in the year 2012.
for Medical Applications imaging segment in Delhi/NCR region
Strategic corporate developments
ƒƒ Digital data remain same after is also an interesting development in
in recent past have further strengthened
multiple copying. this niche marketplace.
the corporate outlook and expectation
ƒƒ It enables displaying images Conclusion
towards this promising market. Some
immediately after acquiring. of the noticeable ones being TowerJazz Growth of diagnostic imaging
ƒƒ Image quality can be enhanced and Panasonic JV in April 2014, equity in India will serve the interest of all
and compared quickly, making funding to CMOSIS by TA Associates stakeholders as introduction of new and
interpretation easier. and China-based Hua Capital’s US$ cost-effective imaging solutions would
ƒƒ It provides a set of images 1.7 billion bid to buy OmniVision be on priority to translate the benefits
for educating the medical Technologies. Investments are not the to the patients. Diagnostic imaging
professionals. only indications towards the prospects sector in India is largely concentrated
3. Global Scenario of Medical and potential this segment offers but in the urban market In future, a major
26 u u



percentage of healthcare facilities their precious money incurred on costly dacryocystogram-nuclear

in small cities and rural areas along tests but also get reduced exposure ƒƒ http://health.economictimes.indiatimes.
with government hospitals will drive from unnecessary radiation. Further, co m / n ew s / d i a g n o st i c s / d i a g n o st i c -
the demand for imaging devices in the India companies should also foray imaging-market-a-window-of-
country. opportunity/46578863
into this growing market. They can
With this growth, it is necessary ƒƒ
leverage Government of India’s ‘Make in
that doctors follow proper ethics and India’, ‘Digital India’ and ‘Startup India’ imaging
such interventions are used only when programmes for the same. ƒƒ
required. There are enough evidences
References: resources/teaching-medical-physics/
in India and abroad that show Doctors gamma/page_54689.html
ordering for medical imaging even astronomy/Detectors ƒƒ
when it is not needed. Therefore, ƒƒ Medicine/
Medical Council of India (MCI) need ƒƒ ƒƒ
to issue strong code of conduct in this lamp cardiology/
regard so that patients not only save ƒƒ ht t p s : / / www.h ea l th ta p .com / top i cs / n

Computer Society of India


Advertisement for EXECUTIVE SECRETARY, CSI at Mumbai

Computer Society of India (CSI), is the largest and oldest The applicant should also be required to travel in
body of IT professionals in India. Established in 1965, with different parts of India, with flair for marketing and minimum
head quarters at Mumbai, CSI has about 18000 Individual 5 years of experience of working with academics/industry
Members and more than 100000 student volunteers. environment.
Computer Society of India is looking to fill up the Executive The application must contain a detailed CV, highlighting
Secretary position at Mumbai. suitability for the position applied for, salary expected with
Executive Secretary: Mumbai two professional references (complete with email and cell
The job involves managing all affairs relating to finances number). Salary will be in commensurate with qualifications
and management of the organization, coordinating with the and experience.
chapters at different locations, managing administration at The applications for the above position must reach
the headquarters, ensuring timely publication and despatch either by hard copy addressed to: Hony. Secretary, Computer
of journals, submission of statutory reports and taxes to the Society of India , Unit No. 3, 4th Floor, Samruddhi Venture
appropriate authorities. etc. and the tasks fixed by ExecCom Park, MIDC, Andheri (E), Mumbai - 400093 or through email:
from time to time, as laid down in the Constitution and Bye- with subject line of the position
laws of Computer Society of India. applied for.
The incumbent should be dynamic with a pleasing Last date for receiving applications is September 20
personality with B Tech / MCA / PG in Technology / (Tuesday) , 2016.
Management with 10-15 years of multicultural experience, For details, please, visit :
less than 55 years of age. Preference will be given to
candidates with experience of working in a similar Prof. A. K. Nayak
environment in a scientific/professional Society. Hony. Secretary

CSI Adhyayan
a tri-monthly publication for students
Articles are invited for July-Sept. 2016 issue of CSI Adhyayan from student members authored as original text. Plagiarism is strictly prohibited.
Besides, the other contents of the magazine shall be Cross word, Brain Teaser, Programming Tips, News Items related to IT etc.
Please note that CSI Adhyayan is a magazine for student members at large and not a research journal for publishing full-fledged research
papers. Therefore, we expect articles should be written for the Bachelor and Master level students of Computer Science and IT and other
related areas. Include a brief biography of Four to Five lines, indicating CSI Membership no., and for each author a high resolution photograph.
Please send your article to
For any kind of information, contact may be made to Dr. Vipin Tyagi via email id
On behalf of CSI Publication Committee
Prof. A. K. Nayak, Chief Editor

u 27 u



A Framework for
Hyperspectral Image Segmentation
B. Ravi Teja K. Venkatarao
Assistant Professor, Dept. of IT, GIT, GU, Visakhapatnam Professor, Dept. of CS&SE, Andhra University, Visakhapatnam

Introduction :
Remote Sensing technology refers physico-chemical mineral properties), Hyperspectral Image
to gather and interpret information Mineralogy (to identify minerals),
Band 1 Band 2 Band i Band z
about an object on the earth without Hydrological science (to determine
any physical contact. This technology changes in wetland characteristics),
uses sensors to construct an image Agriculture (to classify agricultural Dimension Reduction

by capturing the energy emitted and classes) and Military applications Band Selection
reflected from the surface of the earth. (target detection) [2]. An important
There are two kinds of remote sensing task in hyperspectral data processing
systems, passive systems and active is to segment the spectral image Band x Band y Band k

systems. This differentiation is based without losing any valuable details.

on the source of the energy used in Hyperspectral data set consists of Fusion
remote sensing image acquisition. In stack of images from the same scene
passive systems, Sun’s radiation is on the surface of earth. In earlier Clustering Algorithm
used as a source of energy. The sensors studies, to segment hyperspectral
capture the energy which is reflected data set, the algorithm was operated
Segmented Results
by the earth’s surface due to the sun’s on the few selected spectral channels
radiation. This energy measured by the less than five. The efficiency of the
sensor is usually collected in several algorithm depends on the selection
of spectral channels. Some channels Fig. 1: Framework for hyperspectral
spectral bands, with each band having
may contain much information about segmentation
the spectral resolution. Sensors that
collect data of many narrow spectral the scene and some may contain less
bands are called hyperspectral data. information. The performance of any Dimensionality Reduction
In active sensors, the sensor itself segmentation algorithm in terms of The hyperspectral dataset contains
emits energy which is directed towards accuracy will decreases as the number
hundreds of bands with fine spectral
the surface of the earth, reflected by of data channels increases. This high
resolution and spatial information. The
the object and captures the energy by dimensional data will provide data
analysis of this large volume of data
sensor. This article mainly focuses on information content significantly,
causes difficulties in data storage,
but provides a challenge to develop
passive systems, that to hyperspectral processing and transmission. The
algorithms that analyze data more
imaging. dimensionality of hyperspectral data
accurately – A BIG DATA CHALLENGE.
Hyperspectral Imaging In this article, we present a framework strongly affects the performance of any
In hyperspectral imaging systems, to segment the hyperspectral data set segmentation algorithm. Some of the
the sensors are operated from visible by taking the features of all spectral bands in the data set contain redundant
wavelength to infrared wavelength channels. The framework was shown in information and some bands may
capturing simultaneously hundreds Fig. 1. contain less discriminatory information.
of spectral bands (channels) from the The framework for segmentation So it is desirable to reduce the size of
same Earth’s surface area [1]. The data of hyperspectral dataset is done in data set without losing valuable details
collected by sensor is represented in three stages, first- band selection i.e., remove the bands that convey
the form of vector, where each vector (Dimensionality Reduction), second- less information. The band selection
corresponds to specific wavelength. Image fusion of selected bands and methods are classified into two
The hyperspectral images are used third segmentation using supervised/ types, supervised and unsupervised.
in various applications such as un-supervised algorithms. The basic Supervised methods require priori
Ecological science (to study land cover algorithms that are used at each stage information about the object under the
changes), Geological science (to recover are presented in this article. scene, where as unsupervised methods
28 u u



need not require any information. The averaging method of image fusion selection method. These 43 selected
Some of the band selection metrics are is given as: bands are fused into a single image
described below: M
using averaging method of fusion. The
Euclidean Distance [3] is a distance F ( x, y ) = ∑ wk ( x, y ) I k ( x, y )
k =1
fused image is segmented using Fuzzy
measure between any two vectors X and C-means clustering algorithm. The
Y defined as and
result is shown in figure 2.
( X ,Y ) ∑(X k − Yk ) 2 ∑ w ( x, y)=
k =1
i 1, ∀( x, y ) Conclusions
k =1
This article presents a framework
Segmentation for hyperspectral image segmentation.
Spectral Angle Mapper [4] is a angle One of the most important research This work is carried out in three stages-
measure between any two vectors X and topics in hyperspectral data sets is Dimensionality reduction, Image Fusion
Y, defined as segmentation. Segmentation refers and Segmentation. At each and every
X T .Y to partitioning the pixels of the image stage there is a lot of scope to develop
SAM ( X , Y ) = arccos( ) into multiple regions, with each region
new algorithms for efficient analysis
X .Y has specific characteristics. Methods
of hyperspectral data sets. As the
used for segmentation are broadly
Spectral Correlation Mapper [5] is hyperspectral data set is very large,
classified into the following categories:
a correlation measure that measures Histogram methods, Edge detection the problem is considered as a big data
the strength of the linear relationship methods, Region based methods and analysis.
between two vectors X and Y, defined as Clustering algorithms. In this article, References
we mainly focus on Fuzzy c-means 1. lp Erturk,,”Hyperspectral Image
clustering algorithm for segmentation Classification Using Empirical Mode
SAM (X,Y )= of hyperspectral image. The FCM Decomposition with Spectral Gradient
algorithm is described as follows [7]: enhancement”, IEEE Transactions on
Band Correlation [5] is statically a GEoscience and RemoteSensing, Vol 51,
1. Take randomly K initial clusters
correlation measurement that indicates No. 5, May 2013, pp. 2787-2790.
{C1, C2,……,Ck} from the pixels {I1, I2,
the information redundancy of each 2. Ketan
 Kotwal,, “Visualization
spectral band, defined as of Hyperspectral Images Using Bilateral
Nb 2. Initialize membership matrix uij Filtering”,IEEE transactions of Geoscience
S (xip – mi) . (xjp – mj)
with value in range 0 to 1 and value and remote Sensing, Vol 48, No 5, May
BC(i,j) =
Nb Nb
of m=2. 2010, pp 2308-2319.
3. N. Sweet, “The spectral similarity scale
S (xip – mi)2 . S (xjp – mj)2 . Assign each pixel to the cluster Cj and its application to the classification
p=1 p=1
{j=1,2,…..K} if it satisfies the following of hyperspectral remote sensing data,”
Image Fusion of Selected Bands condition [D(. , .) is the Euclidean in Proc.Workshop Adv. Techn. Anal.
After using band selection distance measure between two values]. Remotely Sensed Data,Washington, DC,
methods, we selected some bands uijm D( I i , C j ) < uiqm D( I i , Cq ), q = 1, 2,..., K Oct. 2003, pp. 92–99.
4. B. Park, W. R. Windham, K. C. Lawrence,
from the original data set which contain j≠q and D. P. Smith, “Contaminant
valuable details. These bands are fused classification of poultry hyperspectral
into a single image in-order to produce The new membership and cluster imagery using a spectral angle mapper
a single high contrast image for centroid values as calculated as algorithm,” Biosyst. Eng., vol. 96, no. 3, pp.
visualization and analysis purpose. The 1 323–333, Mar. 2007.
goal of image fusion is to merge all the uik , for1 ≤ i ≤ K
K 1
D(Ci , I k ) m −1 5. O. A. De Carvalho and P. R. Meneses,

features from the selected bands into ∑ (
j =1 D (C j , I k )
“Spectral correlation mapper (SCM):
a single image. Some of image fusion An improvement on the spectral angle
methods such as averaging method,
mapper (SAM),” in Proc. Workshop
∑ uijm I j
Airborne Geoscience, Pasadena, CA, 2000.
1-bit transform method, Principal ^
Cj = n
j =1

Component Analysis, Hierarchical 6. Subhasis Chaudhuri, Ketan Kotwal,

∑ uijm
“Hyperspectral Image Fusion”, Springer
image fusion, Spectral weighted fusion j =1
etc are used for hyperspectral data
1. Continue 2-3 until each pixel 7. J. Harikiran, et al., Multiple Feature
fusion. The averaging method of image
is assigned to the maximum Fuzzy C-means clustering algorithm
fusion is described below [6]:
membership cluster. for segmentation of microarray image”,
The fused image F(x,y) can be International Journal of Electrical and
represented as a linear combination of This basic work is performed on Computer Engineering, vol 5,No 5, October
M selected bands Ik(x,y), with weights Indian Pines data set collected from 2015, pp. 1045-1053, ISSN: 2088-8708.
W for the pixel at location (x,y). The [8]. This data set contains 200 spectral 8.
weights are directly proportional to the bands, with 16 classes. Out of which p h p ? t i t le = H y p e r s p e c t r a l _ R e m o t e _
finer details in the hyperspectral band. 43 bands are selected using ED band Sensing_Scenes

u 29 u



Indiana pines Data set – Image band 100 Fused Image after Band selection Segmented using FCM

Fig. 2 : Hyperspectral image segmentation

Figure 2: Hyperspectral image segmentation

About the Authors:

Dr. K.Venkat Rao is currently working as Professor in the Dept. of Computer Science and Systems Engineering, Andhra University,
K. Venkat Rao He hasis16currently working
years of teaching as His research areas include
experience. Mr. Image
B. Ravi Teja [CSI-I1152582]
Processing, , is working
Big Data, Web Technologies and
other areas. in thebeDept.
He can reached of CSSE, Andhra
at as Asst. Professor, Dept of IT, GIT, GITAM
University, Visakhapatnam. His research University. Currently doing PhD research
areas include Image Processing, Big Data, work in Department of CS&SE, Andhra
Mr. B. Ravi Teja [CSI - I1152582] , is working as Asst.Professor, Dept of IT, GIT, GITAM University. Currently doing PhD research
Web Technologies and other areas. He can University, Visakhapatnam. He can be
work in Department of CS&SE, Andhra University Visakhapatnam. He can be reached at
be reached at reached at

Computer Society of India

Call for Applications
There are various awards which are to be given to the chapters and individuals to acknowledge and encourage for
their extraordinary contribution towards growth of the society and IT industry. Applications are invited from chapters /
individuals for their significant contributions to the society during the financial year 2015-16. The awards will be presented
during the CSI Annual Convention which is scheduled to be held from 8th to 10th December 2016 at Coimbatore.
The CSI awards constituted for encouraging the chapters/ individuals are indicated below:
1. Best National Chapter Award * their nominations in prescribed format which is available
2. Best Regional Chapter Award (Category A) at the above link to CSIED, Chennai at awards@csi-india.
org. Incomplete application will not be considered.
3. Best Regional Chapter Award (Category B)
The nomination should be routed through respective
4. Best Regional Chapter Award (Category C)
Regional Vice President latest by 7th October 2016 to
5. Best Chapter News Letter Award CSI-ED, Chennai and forwarded to “Awards Committee”
6. Chapter Patron Award headed by the Immediate Past President, for final
7. Significant Contribution Award
8. Active Participation Award (Youth) For enquires, kindly contact Mr. Gnanasekaran, Manager
(Administration), CSIED, Chennai at
9. Active Participation Award (Woman)
The norms/ guidelines/ forms for each award are * The best national chapter award will be given on the basis of
available on Chapters and Individuals nomination for best regional chapter award, so, a separate
interested in participating for these awards, may mail nomination for this award is not required.

Prof. Bipin V. Mehta

Chairman, Awards Committee

Mr. Raju L. Kanchibhotla Prof. M. N. Hoda Mr. Ravikiran Mankikar Dr. Suresh Chandra Satapathy
Member Member Member Member
30 u u



Model Based Approach for Effective

Diagnosis of MS-Lesions
K. Naveen Kumar and Y. Srinivas
Dept. of Information Technology, GIT, GITAM University, Visakhapatnam

This article addresses the novel methodology for identification of MS lesions from the brain MRI images.
MS Lesion is a brain deformity which is foreseen during the brain injury or disease. Understanding
these images and identifying the disease is the challenging task. In this article, we address the issue
by proposing a model based Multivariate Generalized Gaussian Mixture Model (MGGMM). The results
derived are analysed with respect to benchmark images and the performance is evaluated by using
quality metrics.

1. Introduction: extracted by considering the images Networks, Support Vector Machines,

The brain consists of three major from Brain web imaging site and results Kernel based approaches and Curvelet
tissues namely white matter, Grey derived are tested against performance based approaches [5,6,7,8]. Models
matter and Celebro Spinal Fluid evaluation using quality metrics such based on generative approaches were
(CSH). Among these tissues, whenever GCE, PRI and VOI. The segmentation also highlighted and these models
a deformity is outseen, it is either process is carried by using k-Means are proven to be more effective than
observed in the white matter or grey algorithm. The rest of the article is the generative approaches [S.K. Pal
matter. As a result of these occlusions, presented as follows. In section 2, a brief and N.R. Pal (1993)]. Therefore, many
the damages in the brain tissues cannot review of the related work is presented of the probabilistic models based on
be outratedly witnessed and thereby and section 3 highlights the model. In Hidden Markov approaches, GMM
causes brain deformities to the brain. section 4 of the article, we deal with with Gibb’s sampling, MRF imaging
The various diseases that generally the experimentation and the results with GMM, Gamma distributions are
focussed among the brain diseases derived are subjected to evaluation and mostly focussed in the literature
are Sclerosis, Lesion, Tumour and In- presented in section 5 of the article. The [10,11,12]. However, some literature is
Homogenities. Identification of each of conclusions together with the scope for also available on Pseudo supervised
these diseases is a challenging task. further improvement are highlighted in models such as clustering algorithms.
Most of the cases that are related the last section of the article. However, as it is obligatory to identify
to brain diseases consist of similar 2. Review of Literature on Brain the damaged pixels more precisely, it is
common symptoms which results in Image Segmentation : therefore necessary to understand the
to misguidance and hence in most of As highlighted in the introduction, pattern of the pixels so that effective
the cases, it results in to improper it is a complicated task to identify the modelling can be subjected and thereby
diagnosis. The impact of these diseases deformities of the brain, therefore effective segmentation can be planned
results in to either Parkinson’s diseases many reviewers have presented their such that the deformities can be more
or Alzimers, Brain dead and in most of ideas by considering different types of precisely rooted out. With this basic
the cases, it leads to vitality. Therefore images such as T1- weighted image, assumption, literature is driven towards
proper identification of the diseases T2-weighted image, Flair and Photon the usage of generative approaches [K.
related to the brain is a challenging density based images [1,2,3,4]. Among Srinivasa Rao, (2007), V. Nagesh et.
task. In this article, we try to address these images, the results derived al.,(2011), GVS Raj kumar,(2011),
the issue of a particular symptom by considering the Flair images are K. Naveen Kumar,(2015)]. Among
and disease – MS Lesion. In order to mostly producing effective results. Lot the generative models available in the
identify the disease, it is customary to of research work has been presented literature, to have better approach, it
segment the medical image and identify off late on Flair images. Many models is customary to identify the pattern of
the deformities. For this purpose, we based on generative, degenerative the pixels and for which multivariate
have developed and designed a medical and Pseudo generative models were approaches have better approximation
image segmentation algorithm based highlighted in the literature. Among the compared to univariate approaches [K.
on multivariate generalised Gaussian non-degenerative models, lot of work is Naveen Kumar, (2015)]. Also the
mixture model. The results are highlighted based on Artificial Neural anatomy of the brain images is mostly

u 31 u



image is gi ( x r ,θ ) and the generalized Gaussian

varying in nature andvector thereforeof the for medical
the distribution of D-
ith iteration.
proper identification of the diseases,
more features are todimension be taken
medical image is g x r , θ and the
in to by generalized Gaussiani distribution
is given

( ) of
T 
∑ ti (x r , θ(l) )
A(N,bij )
(x tij )
(l+1) r =1
consideration before segmentation. m =
 (l) A(N,bij ) (4)
D β K(β ) D-dimension  is givenβ by
xj −µj j   ij T
 j j  ∑ ti (x r---
3. Brain Image Segmentationg(x r / θ) =∏ exp (− A(Dβ j )   bj 
, θ (2) )
2σ= b jK(b jσ ) xj −mj 
jexp (−
 r =1
Model: j=1 j g(x r / θ) ∏ A(b j )  (2)
 2σ j   σ j  Such that A(N, bij ) is some
MS Lesion is a disease that can j=1
 
be witnessed whenever where, µ j , σ j takes
an injury , β j are location, scale and shape parameters.
where, m j , σ j , b j are location, scale and function equal to 1 for b ij = 2 and must
place within the brain regions. There shape parameters. be equal to b − 1 for bij ≠ 1 , in the case
may be several symptoms that are Γ(3 / β j )1/2 of N=2, it is observed precisely that
witnessed during the Alsoinitial
we have K(β j ) =
stages Γ(3 / b j )1/2
A(N, bij ) be an increasing function in
which include headache, neck pain, ΓAlso
(1/ βwe j)
have K(b j ) =
Γ(1/ b j )3/2 terms of bij .
vomiting, vision changes, memory β j /2 b /2
 Γ(3 / β j ) and  Γ(3 / b j )  j
loss, behavioural changes and moving A( b ) =  
and A(β j ) =   withj Γ(⋅)Γ(1 being
/ b ) the gamma function.    3    bij
difficulties. These symptoms are to  j 
   Γ  
 Γ(1 / β j ) 
 
T  (l)   bij   1

be significantly noted and need to be with Γ(⋅) being the gamma function.  ∑ t i (x r , θ )   . x rij − mij bij 
 r =1 
 b Γ 1    
approximated before β jconcluding
≥ 0 is the parameter the that controls the shape of GGD.   ij  b   
disease. Also different types of lesions b j ≥ 0 is the parameter that controls σij(l+1) =   ij   
 T  (5)
the shape of GGD. ∑
 (l)
t i (x r , θ )
are highlighted which damage the brain 
 r =1

and results into different disease such 

as cerebral infraction, multiple sclerosis  
 
and tumour. Therefore, identification of
the proper disease is a critical task. Most 3.3 Initialisation of Model
of the univariate cases fail to generate Parameters:
effective results because of considering The estimation of initial
a single symptom and may result in to parameters is very much needed for
improper diagnosis. Therefore, to have obtaining the revised estimates of the
a more specific identification approach, model parameters. To obtain the initial
in this article, we have considered a estimates, the clustering algorithm is
multivariate approach wherein we utilized since it gives better estimates.
have considered multiple factors such The most generally used method
as memory loss, vision and moving in initialization is to draw a random
difficulty. These features are considered Fig. 1 : Plot of GGM PDF with shape sample within the medical image from
as inputs to the model considered. Figure 1: Plot of GGMparameters, PDF with
b shape parameters,theβentire data set. Using these initial
estimates, the refined estimates are
3.1 Multivariate Gaussian obtained by simultaneously solving the
Generalised Gaussian Mixture 3.2 Estimation of Model Parameters equations (3), (4) and (5) in MATLAB
Model: 3.2 Estimation of Model Parameters Using EM Algorithm:
Using EM Algorithm: environment.
The features considered
In this section, in thethe model Inparameters this section, are estimated the bymodel considering EM algorithm that
above section are taken in to the 4. Experimentation and Results:
parameters are estimated by
proposed model. The cumulative the likelihood function of the model. To obtain the refined estimates In order to depict the proposed
of parameters
considering EM algorithm that
probability density function ofσ each model, we have considered the brain
w i , µij and maximizes the
ij for i=1,2,3,…M; likelihood function
j=1,2,……,D; the expected of the value likelihood or log
of the medical image regions under web images and experimentation is
model. To obtain the refined estimates
conducted by considering brain web
consideration will be of the formfunction is toofbe parameters
likelihood maximized. The wEM algorithm is used for obtaining the refined
i , mij and σij for images. Each of the images is pre-
M estimates. By making use i=1,2,3,…M;
of the stepsj=1,2,……,D;
in EM algorithm, the expected
one can obtainprocessed
the updatedtoequations eliminate the noise and

p(x r / θ) ∑

w i gi x r ,θ ( ) (1)
for the parameters w i , µfunction
value likelihood or log likelihood each of the images are enhanced by
i =1 ij and σij as is depicted
to be maximized. below. The EM using edge enhancement techniques.
 algorithm is used for obtaining the Then the processed images are
where, x r = (x rij ) , is a D dimensional
arbitrary vector for j=1,2,3...D values, refined estimates. By making use of considered for the experimentation.
representing the feature vector; i = the steps in EM algorithm, one can Each image is extracted from the MRI
1,2,..M represents the regions in the obtain the updated equations for the scan images and each of the images is
medical image and the pixels are parameters w i , mij and σij as depicted normalized to 468X468 pixels. The step
represented with i=1,2,3..T values. The below. by step process of the experimentation
parametric set is θ = (m, σ, b) , wi is the T  w (l) .g (x , θ(l) )
  is presented in the segmentation
1 i r
mixing weight such that ∑ w i =1 and the w i(l+1) = ∑  Mi   (3) algorithm in 4.1.
T r =1  w (l) .gi (x r , θ(l) ) 
i =1
probability of i pixel belonging to the
th  ∑ i =1 i 4.1 Segmentation Algorithm:
medical texture feature vector of the ( )
where θ(l) =mij(l) , σij(l) are the estimates at The various steps involved in
32 u u



segmenting the Brain MRI images for Description Original Images Segmented Images
analysis of data to extract meaningful
information for medical diagnosis is
shown below.
Step 1: Consider the brain images
normalized to sizes of 468X468.
Image 1
Step 2: Pre-process the images to
eliminate noise.
Step 3: Enhance the images using edge
enhancement technique.
Step 4: Consider the k-Means algorithm
and segment the images.
Step 5: In order to have effective Image 2
segmentation, consider the features
presented in section 3.
Step 6: The segmentation process is
based on likelihood estimate.
Step 7: Evaluate each segment against
the quality metrics GCE, PRI and VOI.
Image 3
4.2 Results:
The segmentation procedure is
carried out where each pixel from
the image is segmented basing on
the maximum likelihood criteria. The
resultant images are as shown in Fig. 2.
5. Performance Evaluation:
A random sample of five images is Image 4
considered from the Brain web images
database and experimentation is
conducted to evaluate the performance
measures namely GCE, PRI and VOI
and compared with that of Gaussian
mixture model. The image performance
measures viz., GCE, PRI and VOI are
computed for the proposed model. The Image 5
GCE metric formulated by Martin. D et
al. (2001), PRI formulated and given by
Unnikrishnan. R et al. (2007) and VOI
given by Meila. M (2007) are computed
and depicted in the table 1. Fig. 2 : Segmentation Results of Brain Image segmentation

6. Conclusion and Scope for Further

Description Model PRI GCE VOI
The present article presents a Optimal Values 1 0 Infinity
innovative methodology for proper GMM 0.739 0.187 1.196
Image 1
identification of the deformities in the MGGMM 0.884 0.132 0.981
brain and in particular about lesions. GMM 0.815 0.15 1.72
The developed work has a potential Image 2
MGGMM 0.825 0.138 1.648
usage in the field of medical domain
which helps for better treatment of GMM 0.834 0.231 1.02
Image 3
the disease. The results derived from MGGMM 0.984 0.182 0.974
this method showcases that it has GMM 0.812 0.468 1.92
achieved good segmentation accuracy Image 4
MGGMM 0.781 0.444 1.768
with respect to segmentation quality
GMM 0.636 0.364 1.192
metrics. The method is considered Image 5
basing on multivariate features and is MGGMM 0.788 0.331 1.161
addressed for a particular disease of Table 1: Segmentation performance Measures of the Brain MRI Images
brain. However, the methodology can
u 33 u



be further extended to identify the in- similarities and kernel-based centered K. and Srinivasa Rao P.. Article:
homogenities within the brain so that alignment,”  Bio-inspired Intelligence Image Segmentation and Retrievals
better prognosis can be gained. (IWOBI), 2014 International Work based on Finite Doubly Truncated
Conference on, Liberia, 2014, pp. 149- Bivariate Gaussian Mixture Model and
7. References: 153.
1. Ping-Feng Chen, R. G. Steen, A. Yezzi K-Means.  International Journal of
8. V. V. Kumar Raju and M. P. Kumar,
and H. Krim, “Brain MRI T1-Map and “Denoising of MRI and X-Ray images Computer Applications  25(4):5-13, July
T1-weighted image segmentation in using dual tree complex wavelet and 2011. 
a variational framework,”  2009 IEEE Curvelet transforms,” Communications 16. Allili, M.S., Bouguila, N. and Ziou, D.
International Conference on Acoustics, and Signal Processing (ICCSP), (2008), “Finite general Gaussian mixture
Speech and Signal Processing, Taipei, 2014 International Conference on, modelling and application to image
2009, pp. 417-420. Melmaruvathur, 2014, pp. 1844-1848. and video foreground segmentation”,
2. W. Dou, A. Dong, P. Chi, S. Li and 9. Pal, N.R. and Pal, S.K (1993), “A review Journal of Electronic Imaging,  Vol.
J. M. Constans, “Brain Tumor on image segmentation techniques”,
17(1), pp.013005-013013.
Segmentation through Data Fusion Pattern recognition letters, Vol.26(9),
of T2-Weighted Image and MR 17. Naveen Kumar .K, Srinivasa Rao.K,
Spectroscopy,”  Bioinformatics and 10. F. Lavigne, C. Collet and J. P. Armspach, Srinivas.Y and Satyanarayana. Ch(2015),
Biomedical Engineering, (iCBBE) “3D+t Brain MRI segmentation using “Texture Segmentation based on
2011 5th International Conference on, robust 4D Hidden Markov Chain,” 2014 multivariate generalized Gaussian
Wuhan, 2011, pp. 1-4. 36th Annual International Conference mixture model” , CMES Journal,
3. R. Khalaf, J. E. Higgins, R. Roy, P. of the IEEE Engineering in Medicine Vol.107(3),pp.201-221.
van der Zee, L. Dixon and A. Davies, and Biology Society, Chicago, IL, 2014, 18. C.A. Cocosco, V. Kollokian, R.K.-S.
“Image reconstruction for optical pp. 4715-4718. Kwan, A.C. Evans :  “BrainWeb: Online
tomography using photon density 11. S. Liu, J. A. Levine, P. T. Bremer and
waves,”  Biomedical Applications of Interface to a 3D MRI Simulated Brain
V. Pascucci, “Gaussian mixture model
Photonics (Digest No. 1997/124), IEE based volume visualization,”  Large Database”, NeuroImage, vol.5, no.4,
Colloquium on, London, 1997, pp. 1/1-1/7. Data Analysis and Visualization (LDAV), part 2/4, S425, 1997.
4. K. Van Leemput, F. Maes, D. 2012 IEEE Symposium on, Seattle, WA, 19. Martin. D., Fowlkes. C., Tal. D. and
Vandeurmeulen, and P. Suetens: 2012, pp. 73-77. Malik. J. (2001): “A Database of Human
Automatedmodel-based tissue 12. J. Monaco et al., “Image segmentation Segmented Natural Images and its
classification of MR images of the with implicit color standardization Application to Evaluating Segmentation
brain, IEEE Trans. Med. Imag., 18(10): using cascaded EM: Detection of Algorithms and Measuring Ecological
897–908, Oct. 1999. myelodysplastic syndromes,”  2012
Statistics”, Proceedings of 8th
5. N. H. Rajini and R. Bhavani, 9th IEEE International Symposium on
“Classification of MRI brain images International Conference on Computer
Biomedical Imaging (ISBI), Barcelona,
using k-nearest neighbor and artificial 2012, pp. 740-743. Vision, Vol. 2, pp. 416-423
neural network,”  Recent Trends in 13. Srinivasa Rao.K and Srinivas. Y (2007), 20. Unnikrishnan, R., Pantofaru, C.
Information Technology (ICRTIT), 2011 “Unsupervised image segmentation and Hebert, M (2005), “A measure
International Conference on, Chennai, using finite doubly truncated Gaussian for objective evaluation of image
Tamil Nadu, 2011, pp. 563-568. mixture model and hierarchical segmentation algorithms”, Proceedings
6. N. Abdullah, U. K. Ngah and S. A. Aziz, clustering”, Journal of Current of IEEE Computer Society Conference
“Image classification of brain MRI using Science,Vol.93(4), pp.507-514.
on  Computer Vision and Pattern
support vector machine,”  2011 IEEE 14. Nagesh Vadaparthi, Srinivas Yerramalle,
Recognition-Workshops, pp. 34-34.
International Conference on Imaging and Suresh Varma.P: Unsupervised
Systems and Techniques, Penang, Medical Image Segmentation on Brain 21. Meila. M(2007), “Comparing clusterings
2011, pp. 242-247. MRI images using Skew Gaussian - an information based distance”,
7. M. Álvarez-Meza, D. Cárdenas-Peña Distribution, IEEE – ICRTIT 2011, Journal of Multivariate Analysis, Vol.
and G. Castellanos-Domínguez, pp.1293 – 1297 98,pp. 873 –895.
“MRI discrimination by inter-slice 15. Rajkumar G.V.S., Srinivasa Rao n

About the Authors:

Mr. K. Naveen Kumar (CSI-L01141178) is currently Assistant Professor in the Department of the Information
Technology, GIT, GITAM University, Visakhapatnam. His research interests include Image processing, network
security and allied areas. He published good number of papers in journals of repute. He can be reached at /

Dr. Y. Srinivas [CSI-F8001441) is currently Professor and Head of the Department, Department of the
Information Technology, GIT, GITAM University, Visakhapatnam. He had guided more than 10 Ph.D’s in areas of
image processing, software engineering and published papers in Scopus/SCI indexed journals. He also possess
good number of projects funded by UGC, DST etc. He can be reached at
34 u u



Engineering Education In India –

A Roadmap to the Future
N. J. Rao
Vice-Chancellor, Jaypee University of Engineering and Technology, Raghogarh, Guna - MP

1. Introduction in creating jobs, difficult and rather poor into practice through good curriculum
The Indian higher education scene engineering curriculum not meeting and teaching – learning process, We are
has gone through a dramatic change the industry needs. Globalization and heading for a technology driven society
in last 1-2 decades with a phenomenal off shore outsourcing, demand for to address to societal issues to bring
increase in engineering institutions, engineering graduates to be updated, sustainability , security and equity with
particularly with private initiatives. flexible and communicative. These are wellbeing as core trait.
It has resulted in several issues and some of the issues. This is compounded The need is for an engineering
challenges as can be seen from the by rather slow and poor response of student to be adaptable to varying
recent report on engineering education. government and academic institutions. cultures, knowledge levels and
Briefly these are – The key challenges in India are poor markets. While delivering the products
ƒƒ Ph.D and PG faculty is very small in proficiency in language, acute shortage and services, engineer is required to be
Indian engineering institutions, of quality faculty, lack of absorbing a ‘Global Citizen’ meeting the specific
ƒƒ Engineering students going for PG capacity of students, lesser exposure local needs with global knowledge.
courses are small. It is dismal when to reality, rapid growth in engineering How to train the engineer is the biggest
it comes to Ph.D programmes. institutions and number of students. challenge .
ƒƒ Indian engineering institutions are 2. The Challenges The convergence of Info-Bio-
small in terms of size, student This is a period of fast development Nano technologies needed for complex
enrollment, and faculty. globally with great change. The mega systems forces engineers to
ƒƒ Curricula is obsolete and exponential growth of knowledge, be trained in multi-disciplinary traits,
employability of almost 25% increasingly connected global society a path completely different from
graduate is not up to the mark. through information and communication current discipline specific engineering
ƒƒ Diversity of governance systems technology is bringing changes fast. curricula. The need will be to look
and too much controls are Several issues are questioning the old for innovative solutions to approach
mindboggling. values and society is in a great flux in economic, social, security and human
ƒƒ Quality of faculty and shortage is the name of modernism. wellbeing issues. This calls for an input
resulting in poor teaching , high This is coupled with increasing in engineering which is “Holistic” and
failure rates and poor employability. global population, particularly an approach away from traditional. Can
Quality and quantity of technical in developing nations like India. we bring such changes?
education in a country is proportional to Sustainability is at stake. In a knowledge Has engineering moved away
GDP which in turn depends on Human driven well connected society, the from closeness to people and society
Capital, Population, Natural capital / search for skill and resources is global. compared to other professions like
resources, and Investment capital or on The buzz words are “Outsourcing”, and law and medicine? The approach of
per capita income. “off shoring”. The thrust is on market traditional economics based decision
With a staggering 900 million driven economy and this is guiding making has lead to lower priorities to
expected to be in working age group by public policies. The need is open societal, environmental and human
2050 in India, we look at this potential to sources and transparent processes. wellbeing issues. Engineers have
get a competitive edge in the community Free market driven philosophies decide become commodities which can be
of nations through proper competence investments. Result is increasing disposed off when they become obsolete
and skill building. The reality today is regional imbalances and tension, social and do not respond to changing needs.
India fares badly in UN Innovation Index, divide and disparity in society. Indeed the fast pace of knowledge
which depends on Human Capital Index The environment promotes growth makes engineering professional
and Technical capability index. competitiveness and this drives obsolete fast. The need is to wake up and
The engineering courses are failing technology growth. All these need to address to these issues in engineering
to attract large number of students due aim at “Human wellbeing”. The need education to make it relevant and
to several reasons. These include better is to create academic systems and responsive to current societal demands.
job opportunities in other sectors, organizations which capture these The world is passing through
relative neglect of manufacturing sector thought processes and translate them a transition. High expectations

u 35 u



and resource crunch are realities. There are certain harsh realities of ƒƒ Ensure diversity, quality and details
Increasing global tension and societal the developmental process like global are embedded.
divide forces us to revisit engineering warming, climate change, loss of bio-
We need to transform engineering
profession of today to make it relevant diversity, loss of cultural diversity,
education to capture needs of
to 21st century needs. We need changes in moral and ethical values,
tomorrow. The changes taking place
engineering professionals who are increasing life expectancy. The need is
in such spheres like demographics,
competent, dynamic and updated to to ensure that engineers of tomorrow
globalization, society have to be
provide innovative solutions to global must have the skill and knowledge to
adequately captured. It should help
issues. The single minded purpose supplement their technical, analytical
move from natural capital to human
has to be “Human wellbeing” as an strengths with several soft skills
capital to wellbeing.
end result with assured sustainability. related to social, environmental,
The employability of engineering
The curriculum has to develop a new cultural, ethical, moral, communicative
graduates is of prime concern today
mindset in building engineers who and economic domains. We need
which dictates its popularity. With
are capable of sustaining nature, engineers who are ‘Honest Brokers’
relatively low growth in new job creation
conserving natural resources and and true “Global Citizen Engineers”.
and increasing productivity, the situation
creating contentment and wellbeing The engineers of tomorrow need to have
has becomes gloomy. The engineering
among people. We have to move away more innovative skills with technical
education has to focus on this aspect
from the concept of ‘Control nature’ to knowledge, social bent of mind and
by bringing greater industry interaction,
‘Adopt and live with nature’. Technology sustainability approach embedded in
introduction of relevant industry specific
should drive the concept of ‘With nature’ them with the only target of “Human
subjects, creation of greater skills and
rather than controlling it. This paradigm wellbeing” as end result. Innovation
flexibility besides looking at improved
shift is the essence of finding new path is a continuous process of search for
communication skills among graduates.
ways to engineering education. alternate paths. Research provides us
The global challenges dictate
New ways will be found with this the new knowledge to help look at new
the way society conducts itself. The
approach and it will usher in a new products, services , processes, path
challenges include globalization of
world where sustainability will be the ways which ensure capturing the triple
economy, demographics, technological
core motto with contended society and bottom line of 3P’s.
changes, plug and play generation with
economic viability as end results. Indeed Our institutions, most of them ill
a great challenge for engineering equipped, with outdated syllabus, below
profession. Engineering profession par faculty resource, marginalized 3. The Drivers
must move on the ‘Triple bottom line’ in terms of industry interaction and The drivers moving the society are
(namely 3P’s – Planet, People and Profit). policy support of governments, have an ƒƒ Knowledge Economy
Capture the societal and environmental uphill task to take up the challenges in ƒƒ Globalization
issues along with economic viability engineering education. They are to be
ƒƒ Demographics
in engineering decision making. restructured, supported and syllabus
Appropriate technology will provide updated. Practice and theory must ƒƒ Technological Change
the right leverages. This will be the be well integrated in a flexible multi ƒƒ Technological Innovation
corner stone of engineering education disciplinary mode with student centric a) The Knowledge Economy
to address to the challenges. We need focus. The engineer of tomorrow, to
to take well calculated, planned risks be a global citizen engineer, has to Products to Ideas

in altering nature to ensure human be technically competent, globally Manufacturing to services

20th C 21st C
welfare with sustainability. exposed, culturally adaptable, has skills Public policy to markets

The new paradigm of engineering to take calculated risks and innovate, A Knowledge Economy Monopoly to Innovation
education will be based on holistic has entrepreneurial ability. The name
approach which will ensure three of the game is “Flexibility” in teaching We are in an era of abrupt changes,
fundamentals, namely (i) Participate engineering. discontinuities in the moral, ethical fabric
with nature and not against it, (ii) ensure The engineering education process and seeing rapid social transformation.
sustainability and account for ‘Natural must fulfill the following: This is an era where technology
capital’, (iii) Ensure societal welfare and ƒƒ Capture inter disciplinary holistic dominates. Knowledge based society
equity. approach of social, economic is emerging and moving away from
The new education recipe for environmental, legal and local agrarian and industrial era. The time of
engineering must capture the principles issues in global perspectives material and labour intensive days are
of sustainable development through ƒƒ Develop and practice relevant and over and the knowledge intensity will
appropriate technology and resource new technologies with human well drive the products and services. The
conservation and management. The being target advent of internet, social media along
education for engineers must focus ƒƒ Combine tera-conventional-nano with standardization dictate the way
on real world issues of tomorrow scale with info-bio-nano systems. products are managed, developed, the
namely increasing population, limited ƒƒ Capture the mega trends of way production and distribution process
resources in a finite world, increasing digitization, sustainability and manifest. The approach is global and
aspirations of people demanding equity. changing world order solution is local friendly.
36 u u



b) Globalization conventional scale issue (with emphasis drive change should be the motto.
A global economy on close cycle economic models using Lifelong learning should be the dream,
Rich vs. poor the concepts of green, clean and closed and excellent communications skill a
20th C 21st C
Global Resources (oil, cycle operations), nano scale issues (to pre-requisite.
Global sustainability
ensure surface area based properties While technical skill is a must,
of materials and bio systems are interdisciplinary approach to systematic
The impact of globalization exploited). Similarly adoption of digital problem solving is essential. The
on economy expresses in terms of technology is the clarion call of the day. engineer should understand “Multi
smart and capable people advancing e-commerce, e-governance, e-learning, Stake Holder” “Multi objective” decision
technology. Any issue is a global issue e-everything will be common activities making and should be an agent to create
and internet creates transparency and in society. We need interfacing, optimal solutions to societal issues.
democracy. There has to be investment interlinking or cyber infrastructure for Global perspective, ability to innovate
in knowledge resources to ensure the knowledge society to survive. They and integration of knowledge across
economic prosperity and social well will be the new disruptive technologies disciplines is the target for tomorrow’s
being. The need is for highly capable, which will try to provide solutions to engineers.
knowledgeable and skilled workforce global issues and ensure sustainable The decision making process will
that can innovate, apply knowledge. The growth. be built on the principles of 3 SPERM
need is creation of human capital with e) Technological innovation (or life) :
skills, competence and entrepreneurial
Society Security Sustainability
strengths. Can engineering education Commodities to innovation

respond to this need? Analysis to synthesis Profit Production Productivity

20th C 21st C
Reductionism to
c) Demographics consillience Energy Environment Efficiency
Technological Innovation
Aging societies Hierarchy to networks
Resource Recovery Recycle
The global teenager
20th C 21st C Innovation, doing things differently, Material Money Manpower
Population mobility

Demographics Cultural diversity

will be the parameter for success These form the cornerstones of
in the current century. Innovation or modern engineering education.
The global demographic trends are Immovation (Imitation + Innovation)
5. The Plug and Play Generation
fascinating. While developed nations will thrive in education and the result
We have moved from oral to
are aging rapidly, the developing will be productivity improvement,
written words to images (films) to TV to
nations, including India has a significant improvement in standards of living,
Computer to Smart Mobiles. The social
proportion of population below 20 years greater transparency and trust or in
aspect probably did not go hand in hand.
of age. Thus the future of knowledge short Human well being. Technological
From write to read to listen to view is
driven economy is clearly in the hands changes will drive not only new
a transition culminating in multimedia.
of the developing nations. This means technology creation, but innovation,
Traditional class rooms are being
technical education in developing decentralization and democratization.
challenged. Probably gone are the days of
nations like India needs immediate The potential is immense. It has to find
blackboard, chalk and duster. Students
attention. The focus should be on a proper place in engineering education.
are born with a phone in the ears or a
paradigm shift in engineering education The result will be Human welfare
mobile picture in front. They are born
to address to social disparities and with sustainable growth.
in a digital world. They are immersed
regional conflicts on one hand, creating 4. The Future Engineers in videogames, home computers.
human well being through sustainable The future engineers have to be smart phones, instant messaging and
development on the other. “Global Citizen Engineer” and “Honest photographing. The demand of young
We must address to this diversity Broker” at heart. This will come from generation is “Interaction”. They do not
in demographic, namely diversity of their training and learning. This process like one sided lectures. They approach
race, culture, region, economic level, to has to be multidisciplinary flexible learning as “Plug and Play” experience.
create quality education base. learning process. While fundamentals Sequential learning is not in their blood.
d) Technological Change are important, skills in communication, Reading manuals is a waste of time to
team work, adaptation, social and this generation. Instead they like to get
Exponentiating technologies
environmental consciousness are immersed and learn by being a part of it.
20th C 21st C
Info-bio-nano convergence
essential for success. Adaptability This approach needs to be introduced.
Disruptive technologies
should be the passion, and ability to Our teachers need to be reoriented
Technology to social change
Technological Change

Economic Competitiveness
IT, Bio, Nano technologies are New Knowledge (Research) Security
seeing exponential growth. Similarly Economic Competitiveness Social wellbeing & Health
developments opportunities are in Human capital(Education) Sustainability
Tera scale issues (related to ecology, Security Emerging Technologies
energy, infrastructure, healthcare, Infrastructure Policies Inter Disciplinary Activities
logistics, transportation, communication) Large Scale Complex Systems

u 37 u



for the new generation of students. f) 

Research and practical exposure problems and are truly “Honest Brokers”
The current generation of students should become integral to and “Global Citizen Engineers”
believe in “Multitasking” rather than engineering learning process Bibliography:
conventional sequential modes. The 7. Roadmap 1. Azapagic, A., Perdan, S., & Clift, R.
young generation has developed the The challenges to engineering (2004). Sustainable Development in
skill of rapid context changing. This is a teaching learning process are the result Practice – case Studies for Engineers
new challenge for teaching faculty. and Scientists. England: John Wiley and
of human evolution and development Sons.
6. Student-Faculty Learning Teams process. The discipline specific teaching 2. Beteille . A”Universities at the
Interaction, collaboration and adopted so far is considered narrow and Crossroad” - Oxford University Press.
teamwork will be the new mantras inflexible. The multidisciplinary and New Delhi, (2010)
for today’s students who are active sustainability activities are not focused. 3. Biswas, G., Chopra, K. L., Jha, C. S.,
learners. The sequential learning The impact on society has missed & Singh, D. V. Profile of Engineering
of yester years is giving way to non- attention. The new roadmap has to look Education in India: Status, Concerns
and Recommendations. Narosa
linear interactive mode. Multitasking at
Publishing House. . (2010).
and context switching will be the new 1.  Introduction of flexible teaching 4. Duderstadt, J.J. “Engineering for a
order. This needs flexible, problem learning process. Changing World” - Millennium project,
solving systematic approach with hands 2.  Adoption of interdisciplinary Univ. of Michigan http://milproj.
on exposure to problem solving. The approach with stress on knowledge (2008).
active learning student is a member of creation in holistic manner. 5. Manahan, S. E. (2007). Environmental
teaching-learning process, along with 3.  Development of interactive mode Science and Technology – Sustainable
Approach to Green Science. Boca
the teacher. The need is to experience of teaching learning with problem
Raton: Taylor and Francis.
learning and analyse the outcomes, solving approach. 6. Marquita, K. (2004). Understanding
before next step. Technological 4. Sustainability, environmental, Environmental Pollution. Cambridge
Innovation, engineering design will societal and economic issues must University Press (2nd Ed.).
be integral to skill development. That form parts of multi stakeholders 7. McKinney, M. L., Schoch, R. M., &
will be real challenge to engineering - multi criteria decision making Yonavjak, L. (2007). Environmental
institutions. Creativity has to be process. Science – Systems and Solutions.
encouraged. 5.  Stress on communication skills, Boston: Jones and Bartlett Publishers.
8. Rao, N. J. “Challenges before
The new paradigm of engineering morals and ethics and transparency
Engineering Education- Role of
education will include the following : must improve. Humanities and Social Sciences”-
a)  Interdisciplinary teaching with 6. Focus must change to globalized Pertanika Journal of Social Science and
stress on socio, economic, environ- knowledge for localized solutions. Humanities, Vol. 20, No. 4, Dec. (2012),
mental issues for wholesome sus- 7. We need to redefine role of industry, pp. 1249-1280
tainable development. government and professional 9. Rao, P.R. “Higher Technical Education
b) Adopt multi-stake holder – multi- societies in teaching – learning in India- Prospects, Challenges and
Way Forward” - INSA Public lecture,
objective system based problem process.
New Delhi, Feb. 14, (2013)
solving approach with hands on 8. Assessment process must capture 10. Soares, C. M. (1999). Environmental
exposure innovation, doing differently from Technology and Economics – Sustain-
c)  Inculcate in students the spirit rote memorising. able Development in Industry. Butter-
of adventure, calculated risk and 9.  Syllabus must be relevant and worth, Heinemann, Massachusetts.
innovations through technology updated. 11. UNDP Human Development Reports.
d) Create the desired knowledge and 10. Teacher development must be inte- (2010). Environmental Science – The
skills keeping in mind the pace of gral to teaching learning process. Way the World Works 2009 (4th edition)
12. “International Comparative Study:
change and societal needs The roadmap must lead to creation
Engineering Education in India”, Report
e) Imbibe in students lifelong learning of engineering pool who are skilled, No. 57, International Asia Human
habit with focus on transparency, competent, capable of providing Development Sector - April (2013)
responsibility and sincerity holistic systematic solutions to societal n

About the Author

Prof. N. J. Rao is currently Vice-Chancellor of Jaypee University of Engineering and Technology, Raghogarh, Guna (MP). A
graduate/PG from IITKGP, he served over 34 years at IIT Roorkee / University of Roorkee at Chemical Engineering Department
and Department of Paper Technology. He worked for several years as Director of Institute of Paper Technology. He was the
Director of Central Pulp and Paper Research Institute, a national laboratory for one year. He has several awards for best
papers and best teacher and is associated with several National and International bodies like UNEP(NIEM), CPCB, NPC,
MOEF, DST, CSE, HNL, CSE, WBCSD, IL & FS and has visited many countries like China, France, UK, Norway, Sweden,
Finland, Germany, Canada, Thailand, Vietnam, Indonesia. He had been a visiting professor at NTH Trondheim (Norway). An
active researcher, a forceful speaker and an excellent teacher Prof. Rao had several publications to his credit. He can be
reached at
38 u u


Computer Society of India
Academic Awards 2016
Call for Applications
Computer Society of India has been honouring academic excellence through Academic Awards every year. The awards will
be presented during the CSI Annual Convention which is scheduled to be held from 8th to 10th December 2016 at Coimbatore.
Applications are invited for the following awards for the period from July 2015 to June 2016 from the accredited student
branches who meet the criteria and are currently in good standing.

S. Name of the Award Criteria To be submitted by

1 Best Accredited Student Branch Good standing – during the award year Student Branch Counsellor (SBC)
Award and currently, large student strength & with necessary recommendation
large number of activities as defined in from Regional Student’s Coordinator
the specified form (RSC) and approval from Regional
Vice President (RVP)
2 Largest Student Branch Award Continuous good standing for the past Decided by Awards Committee
3 years with highest 3 years averaged
3 Best CSI International Students Institutional member hosted maximum SBC with necessary recommendation
Event Host Award students competition participated by from RSC and approval from RVP
minimum 10 foreign students
4 Highest Sponsorship of CSI Institutional member extending SBC with necessary recommendation
Events Award maximum support for CSI events during from RSC and approval from RVP
the award year
5 Longest Continuous SBC Award Longest continuous tenure as SBC over SBC with necessary recommendation
the last 3 years from RSC and approval from RVP
6 Faculty with maximum Publishing maximum articles in CSI Self with necessary recommendation
publishing in CSI Publications publications digital library during the from RSC and approval from RVP
award year
7 Paper Presenter at International Presentation of paper at prestigious Self with necessary recommendation
Conference for Faculty International Conferences during the from RSC and approval from RVP
award year
8 Students with maximum Publishing maximum articles in CSI SBC with necessary recommendation
publishing – CSI publications publications digital library during the from RSC and approval from RVP
award year
9 Highest Committed Accredited Most active CSI Volunteer from the SBC with necessary recommendation
Student Branch Activist Award Student Branch during the award year from RSC and approval from RVP
10 Best Ph D Thesis Award CSI member, who submitted a high- Research Scholar (who got the Ph D
quality thesis (Thesis quality to be during the award year) / the Research
evaluated by a panel of eminent research Supervisor / Current Employer
scientists) leading to acceptance for Ph D
degree by a recognized University
The applications for the awards are invited only from the CSI members or from CSI Accredited CSI Student Branches in good
standing during the current year as well as during the Award year. Application Forms are available at http://www.csi-india.
The applications should reach via email – in specified form latest by 7th October 2016.

Prof. Bipin V Mehta Mr. Raju L Kanchibhotla Prof. M. N. Hoda Mr. Ravikiran Mankikar Dr. S. C. Satapathy
Chairman, Awards Committee Member Member Member Member

u 39 u


Computer Society of India

Regional Student Coordinators for the year 2016-17

Region Name & Affiliations Region Name & Affiliations
I Mr. Saurabh Agrawal V Prof. C. Srinivas
Head, Marketing and TTH Associate Professor in CSE
Delivery (MEA), TCS, Kakatiya Institute of Technology & Science,
Ghaziabad-201010 Warangal, Telangana
Mob : +91-92668 39339 Mob : +91-90300 70247

State Student Coordinators for the year 2016-17

Region Name & Affiliations Region Name & Affiliations
I Prof. M. U. Bokhari V Dr. A. Kanaka Durga
Professor & Chairman in CS Professor in CSE and HOD - IT
Aligarh Muslim University (AMU) Aligarh Stanley College of Engineering and

Technology for Women, Hyderabad


Mob : +919412640294 Mob : +91- 98493 84247

Prof. Amarjeet Singh Prof. Srikanth Chintakindi
Professor in CSE and HOD - IT

Professor in CS

Himachal Pradesh (HP) University, Shimla Stanley College of Engineering and

Mob : +91-94184 84855 Technology for Women, Hyderabad Mob : +91-98493 84247
Dr. Maninder Singh Prof. P E S N Krishna Prasad
Associate Professor in CS & IT Dept. of Computer Science & Engg

Thapar University, Patiala Prasad V Potluri Siddhartha Institute of

Mob : +91-98156 08309 Technology, Kanuru, Vijayawada
Andhra Pradesh Mob : 91- 94949 55850
Dr. Anupam Baliyan
Associate Professor Prof. Pritee Parwekar
Bharati Vidyapeeth’s Institute of Computer CSE Department

Applications and Mgmt. (BVICAM), New Delhi Anil Neerukonda Inst. of Tech. & Sciences.
Mob : +91-98183 27524 Sangivalasa, Vishakapatnam Mob : +91-90002 49712
Dr. Ashutosh Kumar Bhatt

Prof. Suman Jayakumar

Dept. of Computer Science

Asst. Professor, Dept. of ISE,

Birla Inst. of Applied Sc. Bhimtal, Nainital
VVIET, Mysore
Mob : +91-94123 27953
Mob : +91- 99721 37917
Dr Brijesh Kumar VI Prof. Ramrao Surya Wagh
Dean Academics and Dean SOET

Associate Professor in CST

K R Mangalam Univ., Sohna Road, Gurgaon Goa University, Goa
Mob : +91-9811326840

Mob : +91-94238 82964
Dr. Muheet Ahmed Butt Prof. A. R. Anil
PG Dept. of Computer Science

Head & Professor


University of Kashmir, Srinagar (J&K) Sree Budha college of Engineering, Patoor


Mob :+91- 94194 26555 Mob : +91-94474 77577
Dr Jasbir Singh
Jammu (J&K)

Sr. Assistant Professor in CS & IT

University of Jammu, Jammu -180006
Mob : +91-9419115626

Like CSI on facebook at :
40 u u



CrossWord Durgesh Kumar Mishra

Chairman, CSI Division IV Communications
Professor (CSE) and Director Microsoft Innovation Center, Sri Aurobindo Institute of Technology,
Test your knowledge on Medical Image Processing
Solution to the crossword with name of first all correct solution provider(s) will appear in the next issue. Send your answer to CSI
Communications at email address and cc to with subject: Crossword Solution –
CSIC September 2016 Issue.

1 2

1. Having ultrasonic frequencies
3. A biomaterial that is made up of polymer chains
3 4 5. An X-Ray imaging technique to detect breast
6. Two photon absorption
7. A measure of strength of magnetic field
8. An standard related to images
5 9. Mathematician whose theory used in signal
10. Device used to protect against radiation

2. Study and treatment of cancer
4. A small telescope like viewing instrument
8 5. X-Ray of spinal cord
7. X-Ray device movement to view body details

Solution for August 2016 Crossword
1 2
3 4

10 T M E T A B A L L

5 6

7 8


We are overwhelmed by the response and solutions received from our enthusiastic readers P E




All nearby Correct answers to August 2016 month’s crossword received from I Y

the following reader: Dr. Sandhya Arora, Professor, Cummins College of C

Engineering for Women, Pune. 13


Report on Colombo Visit of Dr. Anirban Basu

Dr. Anirban Basu, President CSI visited Colombo from August 8-10 at the invitation of Computer
Society of Sri Lanka (CSSL) to deliver a keynote talk in their Annual Conference NITC 2016. The
conference was held in Galadari Hotel, Colombo and attended by about thousand delegates including
researchers from Australia, Bangladesh, Malaysia, India etc.. The theme of the Conference was
“Enhancing Sri Lanka Through Digital Disruption”. The Conference was inaugurated by Sri Lanka’s
Minister in presence of Prof Ashu Marasinge, Member of Parliament, Sri Lanka. The delegates were
from Sri Lanka and other parts of South East Asia who did not have a clear idea about the advances
India has made in the field of IT and the changes happening in India. In his talk, Dr. Basu gave a
glimpse of India’s financial prowess, diverse culture and several recent initiatives of the Government
of India. He presented the role of Computer Society of India in catalyzing the growth of IT in India before delivering talk on “IoT, Big Data and
Analytics” which was very well appreciated.
Dr. Basu discussed several areas where India and Sri Lanka can work together as the Government of Sri Lanka was also looking at ways for
growth of IT and taking several initiatives like Smart City project etc.. He also interacted with the President of CSSL, Dr Dayan Rajapakse and
the Vice President Mr Yasas Abeywickrama on how CSSL and CSI can work together and how CSI can play a more effective role in SEARCC.

u 41 u




CSI Allahabad chapter conducted an expert lecture on

“Recommender System” by Dr. Pragya Dwivedi on Saturday,
30-07-2016 at 5 PM at MNNIT Allahabad (Venue: Design
Computer Society of India, Coimbatore Chapter
Center Seminar Hall). In this lecture, Dr. Dwivedi focused
organized a seminar on “It’s a ‘polymath’world - success
on recommender system, recommendation techniques are
secrets for an emerging world” on 12.08.2016, 5.00 pm
mainly Collaborative Filtering, content based filtering and
at CSI Hall, Coimbatore. More than 70 participants from
hybrid filtering. She also gave a light on various domains like
various organizations in and around Coimbatore attended
E-learning, multi-criteria and cross domain recommender
the seminar session.
system. The next day of lecture (31-07-2016), chapter
Ms. Devadarshini and Mr. Jelix Johnson from SNS
conducted young talent search programming contest in
College of Engineering (Partner) have addressed the
UPTECH Allahabad where four colleges’ teams of Allahabad
committee members. Mr. Vishnu Potty, Chairman, CSI
were participated.
Coimbatore Chapter welcomed the gathering and also
BHOPAL CHAPTER briefed the session topics of the seminar and introduced the
keynote speaker Mr. Subramaniam to the participants.
The Chief Guest explained about the “Generation of
polymath world” and also he demonstrated the polymath
pattern in the form of wave pattern in three ways such
as Internet connection and their applications, apple, Ios,
apps and Internet of things, automation etc. He explained
in detail structure about the way of Generation changes by
CSI Bhopal Chapter supported two week Faculty the OLA taxi app and he gave us an idea about the android
Development Programme on BIG DATA from 1-10 August development for automation. The session helped the
2016 organized at Sagar Institute of Research and participants to invoke the concept to their future generations.
Technology, Bhopal. The programme was sponsored by The seminar was well received by the participants and
Department of Science and Technology (DST), Govt. of India ended with Q&A session. Finally, vote of thanks was given
and technically supported by CSI Bhopal Chapter and ACM by Dr. G. Radhamani, Secretary, CSI CBE Chapter.
Bhopal Chapter.
Eminent personalities like Dr. P K Chande (Former
Professor IIM Indore) from Indore, Dr. S Sampath from
Madras University, Dr. K K Shukla from IIT Rourkee, Dr. R
Bala Venkat Subramanium from NIT Warangal, Dr. Indranil
Mukherjee from ISI Kolkata, Dr. T Vijay Kumar from JNU
New Delhi, Dr. Parag Kulkarni CEO from Pune, Dr. P K
Singh from IIITM Gwalior, Dr D S Tomer from MANIT Bhopal,
Dr. R S Thakur from MANIT, Bhopal, Dr. Vivek Tiwari from
IIIT Raipur and Dr. S Prajapat from IIPS Indore Mr Sunil
Jain from National Informatics Centre Bhopal and others
from different Established Organization of India graced this CSI Coimbatore Chapter monthly seminar was
programme. organized on 18 July 2016. The meeting started with the
Prof. Rajesh Shukla (Coordinator) and vice president welcome address by Mr. Vishnu Potty, Chairman, CSI,
of CSI Bhopal Chapter enlightened that more than sixty Coimbatore Chapter and introduction of the chief guest.
participants from different corners of India like Hyderabad, Mr.Brijraj Singh addressed the audience and discussed the
Bengaluru, Pune , Jaipur, Chennai and different institutions following:
of Madhya Pradesh participated in this prestigious Evolution of Azure, Open Source Initiative in IAAS,
programme. In this momentous occasion Er Sanjeev Open Source Initiative in PAAS, Open Initiative in Tools and
Agrawal (Chairman SGI), Dr Prashant Jain (Vice-Chairman) frameworks
congratulated all the Dr. Akhilesh Upadhyay (Director SIRT) The seminar was well received by the participants and
focused that this kind of programme will be organized in ended with Q & A session. Finally, vote of thanks was given
future also. by Dr. G Radhamani, Secretary, CSI CBE Chapter.
42 u u



Computer Society of India (CSI), Coimbatore Chapter CSI Hyderabad Chapter conducted Two day Workshop
and Department of MCA, Hindusthan College of Arts on Analytics using R was organized in CSI Hyderabad
and Science, Coimbatore jointly conducted the first level Chapter (Regd) Office Hyderabad during 2-3 July 2016.
regional competition for “CSI Young Talent Search Computer Twenty participants from different organizations such as
Programming” on 31-07-2016 at Hindusthan College of Arts Indian Statistical Institute. Calcutta, INFLIBNET Centre,
and Science. Dr. A. V .Senthilkumar, Director, Dept. of MCA, Gandhinagar Gujarat, NTT Data Hyderabad, S.V.Subba Rao
Hindusthan College of Arts and Science acted as a Chief Chartered Accountant & Co. Hitachi Consulting, Hyderabad,
Superintendent for the examination, Dr. G. Radhamani, Adama Science and Technology University,Ethiopia,
Secretary – CSI-Cbe Chapter, Mr. A.Sivabalan, CSI-Cbe Thrinaina Informatics Ltd. Secundrabad , VNR Vignana
Chapter, Mr. C. Thirumoorthi, Assistant Professor, Dept. of Jyothi Institute of Engg & Technology, Geethanjali College
MCA , Hindusthan college of Arts and Science were Present. of Engg & Technology, Loyola Academy - Degree and PG
More than 20 school students from various parts in and college, Nalla Malla Reddy Engg.College Hyderabad and
around Coimbatore region participated in the competition. Wissen Infotech, Hyderabad
Valedictory was addressed by Shri Sudhakar
HARIDWAR CHAPTER Chairman and Managing Director ECIL Hyderabad, Mr.
Raju Kanchubatla RVP-V CSI India, Mr. Chandra Dasaka,
Director-Mobile Digiconverse Private Limited, briefed about
the Workshop details and Dr DV Ramana Wissen Infotech,
Hyderabad provided the Workshop Summary and Prof
Krishna Prasad AV, KL University Concluded with the Vote
of Thanks. Resource persons from Prof Sudhakar, Raskey
Software Solutions (P) Limited, Osmania University, Indian
Institute of Public Health, Hyderabad and Prof.Rajesh
Prabhakar Kaila, Visiting Faculty, Symbiosis Institute Of
Business Management - Symbiosis International University,
Hyderabad shared their extensive knowledge on the topics
covered during the workshop were: Introduction to Analytics,
On 22nd August, 2016 a cloud computing workshop Descriptive Statistics – Measures of Central Tendency,
was held at Faculty of Engineering & Technology, Gurukula Measures of Dispersion and Measures of Association.
Kandgri Vishwavidyalaya, Haridwar by CSI Haridwar Importance of R and Basics of R Data Manipulation & Data
Chapter. The workshop had total 3 sessions started with the Handling, Data Visualization – I- Line Plots, Bar Plots, Stacked
“Introduction of the Cloud Computing” taken by Abhigyan Bar plots, Pie chart, scatter charts, box plots, histograms,
and Vivek Ji. Later the chief guest of the workshop Mr. heat maps, Tabulation of data, Hypothesis Testing – One
Mani Madhukar, IBM India, talked about the block-chain Sample T test, Two Sample T test, One Way Anova, Multiple
technology and method to implement by using IBM Cloud Linear Regression – Procedure, Model Fitting and Predictive
Bluemix, also made us aware about few other technology Modelling, Excel and R Integration.
GLAN, Importance on node.js in cloud. Later he discussed The Workshop was completed successfully with a good
various issues of cloud computing with the students. The rating and the participants were very much appreciated by
third session was the implementation of IBM insights knowledge sessions and the hospitality.
and Bigsheets taken by Mohit. The whole day was very
informative. Total 103 students took part in the workshop KOLKATA CHAPTER
and on this occasion Dr. Krishna Kumar, Chairman CSI
Hairdwar Chapter, Dr. Mayank Aggrawal, Vice-Chairman
CSI Hairdwar Chapter, Mr. Nishant Kumar, Secretary CSI
Hairdwar Chapter and Mr. Manish Aggrawal, Treasurer CSI
Hairdwar Chapter was present. Organizing secretary for the
workshop was Mr. Nishant Kumar.


CSI, Kolkata Chapter’s Fifth Lecture of Lecture Series

was held on 06.08.2016 at 4.30 pm in the CSI Kolkata Chapter
office. Mr. Supratik Gupta, CIPSP® CLIP, deliverd his expert
lecture on “Technology in Old Age Care”.

u 43 u



University. The program has started with the welcome note

by Prof. Dipali Thakkar, Marwadi University by welcoming
Prof. Sunil Bajeja, Immediate Past Chairman, CSI Rajkot
Chapter and guest faculty Prof. Rajesh Nagawade..

CSI Kolkata Chapter organized one-day course on

“Research Methodology”, on Saturday, 30th July, 2016.
The lectures addressed issues related to motivation and
approach to research. The course has been conducted by
Dr. Dipti Prasad Mukherjee, Professor of Electronics and The vote of thanks was presented by Prof. Jobi Jose,
Communication Sciences Unit, Indian Statistical Institute. Secretary CSI Rajkot Chapter. He has played a key role in
22 participants attended the course arranging a successful seminar and made all arrangements
related to session.
Certificates were distributed to all the participants by
the founder Chairman of the chapter Dr. R. Sridaran.


Eastern Regional Student Convention has been CSI Vellore Chapter organized a one day Workshop
organized on 20th August at MCKVIE institute of Technology, on “Apps for GRE & TOEFL” on 27/7/2016 at VIT University.
Liluah, Hourah. Prof.(Dr.) A. K. Nayak, Secretary, Computer Ms. Pavithra Srinivasan, MS.,Stanford University, USA.
Society of India was the Chief Guest on the occasion. Conducted a diagnostic test for CSI students and explained
Students from various colleges and universities attended how to solve difficult questions, mapping scores to jog
the convention. Sixteen officials/members CSI of Kolkata opportunities, life culture in US, around 90 CSI life members
Chapter attended the conference. Among the activities, attended the workshop, organized by Prof. G. Jagadeesh
twelve papers were presented by the students of various and Prof. K. Govinda
institute like Jadavpur University, Kalyani University etc.
There was quiz contest and extempore competition. More
than 60 students participated in the convention. Regional
meet was also held on the same day where participants Congratulations
from different chapters of this region participated like
Patna, Kolkata etc.
First level student programming contest was held
on 31st July 2016 at BIT Meshra Campus, Kolkata Center.
Seventeen schools were participated in the competition.
Prof. Jayesh M. Solanki, Chairman - CSI Ahmedabad Chapter has
been appointed as a Technical Advisor in Gujarat University for
RAJKOT CHAPTER implementing online centralised admission process. Currently,
Computer Society of India – Rajkot Chapter, has he has been appointed as an Officer on special duty and Co-
organized a two hour Seminar on ’’Introduction to Linux ordinator for BA/MA programs for implementing this project in
most challenging environment. More than one lakh candidates
Programming Environment” on 30 July 2016. Total 25
have been processed through this online portal successfully.He
participants attended this Seminar. received an honour from the Hon’ble Vice chancellor of Gujarat
The session was taken by a very renowned personality, University on August 15, 2016.
Prof. Rajesh Nagawade, Corporate Manager, Marwadi
44 u u



Report Submission
School of Computer Studies, Ahmedabad University

Student branches are requested to send their report to with a copy to
Chapters are requested to send their activity report to
Kindly send high resolution photograph with the report.
Contact Dr. Vipin Tyagi, Editor – CSI Communications at for any query.
30-7-2016 – Prof. Bipin V. Mehta & Dr. Nirav Thakkar during
inaugural session on ICT in Education for Digital India seminar
G H Patel College of Engineering & Technology, Vallabh Vidyanagar

8-8-2016 – Seminar on Tips and Tricks for Competitive Coding 12-8-2016 – Event on Qriosity : A General Quiz
Geethanjali Institute of Science & Technology, Nellore Rajarajeswari College of Engineering, Bengaluru

19-8-2016 – Dr. Ch Pradeep Reddy during Guest Lecture on 19-7-2016 to 25-7-2016 – Faculty Development Program on
IOT and Cloud Computing Bridge the Gap-VTU CBCS Scheme
Anurag Group of Institutions, Hyderabad

30-7-2016 – Workshop on Web Development 3-8-2016 – Orientation program

u 45 u



Vasavi college of Engineering, Hyderabad MLR Institute of Technology, Hyderabad

27-7-2016 - Contest on DIGBUG-DEBUG 16-6-2016 – Awareness program by Mr. Pattabiraman, TCS

CMR Technical Campus, Hyderabad

28-6-2016 - Guest Lecture on Big Data Analytics 28-6-2016 - Guest Lecture on Intellectual Property Rights

Sphoorthy Engineering College, Hyderabad G Pullaiah College of Engineering & Technology, Kurnool

4-8-2016 – Mr. Raju Kanchibhotla, RVP-5 inaugurated the 29-7-2016 – Mr. Anil Kumar, Dr. Nageswara Rao, & Dr. Prem
Student Branch Kumar during Student Branch inauguration
Lendi Institute of Engineering & Technology, Vizianagaram GSSS Institute of Engineering and Technology for Women,

13-8-2016 – Workshop on Web Designing 6-8-2016 – Mr. Girish presenting Technical talk on
Architecture and Structure of DMBS and RDMBS
46 u u



NBKR Institute of Science and Technology, Nellore Malla Reddy Inst. of Technology & Science, Secunderabad

3-8-2016 - One Day Workshop on Presentation Skills 19-7-2016 – Student Branch 3rd Annual Day Celebrations
Vishwakarma Institute of Information Technology, Pune

21-7-2016 & 22-7-2016 - Arduino Workshop 22-7-2016 – Expert Lecture, Best Coding Practices in Data Structure
Institute of Management Research & Development, Zeal Institute of Business Administration, Computer
Shirpur Application and Research, Pune

9-8-2016 – Event on Opportunities in Government Sector for 6-8-2016 – Guests and Participants during National Seminar
Fresh Graduates on Digital Assistance
All India Shree Shivaji Memorial Society’s Institute of Information Technology, Pune

11-8-2016 – One day workshop on Software Testing 12-8-2016 – One day workshop on Latest Trends in Cloud

u 47 u



Knowledge Institute of Technology, Salem Sri Sai Ram Institute of Technology, Chennai

25-7-2016 & 26-7-2016 – Mr. Prateek Gupta during two day 13-7-2016 to 15-7-2016 – Mr Bhaskaran, Past Chairman,
workshop on Internet of Things (IoT) CSI Chennai Chapter speaking on the EDC camp
National Engineering College, Kovilpatti Syed Ammal Engineering College, Ramanathapuram

22-7-2016 – Event on BUG-D (C-debugging contest) 23-7-2016 – One day workshop on Appdhoom-3 -
Internet of Things development
Valliammai Engineering College, Kattankulathur

26-7-2016 – Guest Lecture on Issues in Network Security 29-7-2016 – Dr. Vanathi, Head CSE addressing the gathering
during Seminar on Android Application Development.
VIT University, Vellore Einstein College of Engineering, Tirunelveli

10-8-2016 – One day coding competition on Crackerjack 8-8-2016 – Dr Ramar, Dr Velayutham & Prof Suresh
Thangakrishnan during Motivational Talk
48 u u



Velammal Engineering College, Chennai MEPCO Schlenk Engineering College, Sivakasi

20-7-2016 - Student Branch Inauguration 25-7-2016 - CSI Student Branch Office Bearers interacting
with the Freshers during the event on Quiz
Nehru College of Management, Coimbatore

19 & 20-7-2016 - Two Days Workshop on PHP 29-7-2016 - Industry Institute Interaction

CSI Chapter Elections 2017-2018/2019

Chapter Nomination Committee shall invite nominations for the following positions from their respective
eligible voting members. In case you wish to update your details, please write to CSI HQ for updation.

For the term 2017-2018 (April 1, 2017 - March 31, 2018)

1) Vice Chairman-cum-Chairman Elect- One Post
2) Nomination Committee (3 members)- 3 Posts.
3) Managing Committee:(4 / 6 / 8 members as per class/category of chapter)
Category A – (Chapters having more than 500 members) – 8 MC Members.
Category B – (Chapters having 250 – 500 members) – 6 MC Members.
Category C – (Chapters having less than 250 members) – 4 MC Members.

For the term 2017-2019 (April 1, 2017 – March 31, 2019)

4) Hon. Treasurer - One Post
Chapter election process must be completed by 15th November 2016 i.e. the election slate/ results must be communicated
to CSI HQ at by chapters by this date. If the chapter opts to go for online voting with National ExecCom
elections, all details of the nominees must be sent to CSI-HQ/ National NC by 31st October 2016.The election process in such
cases would be completed by 31st January 2017.
All queries/ doubts pertaining to chapter elections can be mailed to respective Chapter NC. In case of any difficulty you can
also write to National NC at with a copy to CSI-HQ at

National Nomination Committee (2016-2017)

Mr. Ved Parkash Goel Dr. Santosh Kumar Yadav Mr. Sushant Rath
Chairman Member Member

u 49 u


11th INDIACom; 2017 4th IEEE International Conference on

“Computing for Sustainable Global Development”

(01st – 03rd March, 2017)
Organized by
Bharati Vidyapeeth’s Institute of Computer Applications and Management (BVICAM), New Delhi
Technically Sponsored by
IEEE Delhi Section
Supported by
Computer Society of India (CSI), Divisions-I, II, III, IV & V and Region-I,
Institutions of Electronics and Telecommunications Engineers (IETE), Delhi Centre,
Indian Society for Technical Education (ISTE), Delhi Section,
Institution of Engineering and Technology (UK), Delhi Local Networks and
Guru Gobind Singh Indraprastha University (GGSIPU), New Delhi

Paper Submission Deadline: 24th October, 2016 [No Further Extension]

Paper submission Link:

Conference Website:

Announcement and Call for Papers

INDIACom-2017 is aimed to invite original research papers in the field of, primarily, Computer Science and Information
Technology and, generally, all interdisciplinary streams of Engineering Sciences, having central focus on sustainable
computing applications, which may be of use in enhancing the quality of human life and contribute effectively to realize
the nations’ vision of sustainable inclusive development using Computing. INDIACom-2017 will be an amalgamation of four
different Tracks organized parallel to each other, in addition to the 3rd International Workshop on Information Engineering
and Management (IWIEM-2017) and few theme based Special Sessions, as listed below:-
• Track #1: Sustainable Computing
• Track #2: High Performance Computing
• Track #3: High Speed Networking & Information Security
• Track #4: Software Engineering & Emerging Technologies
• Track #5: Theme Based Special Sessions
Instruction for Authors
Authors from across different parts of the world are invited to submit their papers. Authors should submit their papers
online at New authors should first sign up and create an
account on to log in and submit paper. Only electronic submissions will
be considered. Paper submission, as E-Mail attachment, will not be considered.
Important Dates
Submission of Full Length Paper 24 October, 2016
Paper Acceptance Notification 14th January, 2017
Submission of Camera Ready 23rd January, 2017 Registration Deadline (for inclusion of 23rd January, 2017
Copy (CRC) of the Paper Paper in the Proceedings)
Accepted Papers will be published in IEEE Xplore, which is indexed with world’s leading Abstracting & Indexing (A&I)
databases, including ISI, SCOPUS, DBLP, EI-Compendex, Google Scholar, etc. Further details are available at www.bvicam. All correspondences, related to INDIACom-2017, must be addressed to:
Prof. M. N. Hoda
General Chair, INDIACom-2017
Director, Bharati Vidyapeeth’s Institute of Computer Applications and Management (BVICAM)
A-4, Paschim Vihar, Rohtak Road, New Delhi-110063 (INDIA)
Tel.: 011-25275055 • TeleFax: 011-25255056, Mobile : 09212022066
50 u u



Sanjay Mohapatra, Vice President, CSI & Chairman, Conf. Committee, Email:
Date Event Details & Contact Information

SEPTEMBER Workshop on Analytics using R

10-11, 2016 Contact :
16-17, 2016 2016 International Conference on Frontiers of Intelligent Computing: Theory and applications (FICTA), KIIT
University, Bhubneswar. Contact :
OCTOBER National Conference on “Recent Advances in Computer Science & Technology”, RACST-2016, Department
04-05, 2016 of Computer Engineering, G H Patel College of Engineering & Technology, Vallabh Vidyanagar, Gujarat, Contact : Dr. Maulika Patel,
06-08, 2016 International Conference on “Computational Systems and Information Technology for Sustainable
Solution [CSITSS-2016]” Organized by CSE & ISE & MCA - R.V. College of Engineering, Bengaluru -560059.; Contact :; Ph: 080-67178183, 8180;
NOVEMBER Third International Conference on Computer & Communication Technologies (IC3T - 2016) at Devineni
05-06, 2016 Venkata Ramana & Dr. Hima Sekhar MIC College of Technology, Vijayawada, Andhra Pradesh, India.
Contact : Dr. S.C. Satapathy, 9000249712,, Dr. K. Srujan Raju, 91-9246874862, Prof. Vikrant Bhateja, 91-9935483537,
11-12, 2016 International Conference on Advances in Computing and Data Sciences (ICACDS-2016). Proceedings
by Springer CCIS/LNCS Organized by Krishna Engineering College (KEC), Ghaziabad. http://icacds2016. Contact : Dr. Mayank Singh, Mob: 09540201130
National Conference on Smart And Innovative Technologies in Engineering And Sciences (SITES 2016)
Gyan Ganga College of Technology, Jabalpur, MP. Contact sites:
17-19, 2016 Interntional Symposium on Acoustics for Engineering Applications : Acoustics for Quality Improvement in
Life at KIIT, Gurgaon Contact: ​Prof. (Dr.) S. S. Agrawal, Chairman, OC – NSA-
2016, Director General KIIT Group of Colleges, Gurgaon Formerly: Emeritus Scientist CEERI/CSIR,Advisor
CDAC-Noida. Email:
18-20, 2016 2nd International Conference on Communication Control and Intelligent Systems, at GLA University,
Mathura .​ ​ ​Contact: ​
22-25, 2016 Special session on “Smart and Ubiquitous Computing for Vehicle Navigation Systems” at IEEE TENCON
2016, Marina Bay Sands, Singapore (
Contact : Dr. P.K. Gupta, Prof. Dr. S. K. Singh
DECEMBER National Symposium on “Recent Advances in Remote Sensing and GIS with Special Emphasis on Mountain
07-09, 2016 Ecosystems” and their Annual Conventions at Dehradun.
Contact: Dr. S. K. Srivastav, Organising Secretary & Group Head, RSGG, Indian Institute of Remote Sensing,
Indian Space Research Organisation, Department of Space, Government of India , Dehradun, India - 248 001.
email :
08-10, 2016 CSI Annual Convention (CSI-2016): Theme: Digital Connectivity - Social Impact; Organized by CSI Coimbatore
Chapter; Pre-Conference Tutorial on 7th Dec. 2016 Venue: Hotel Le Meridien, Coimbatore
Contact : Dr. Ranga Rajagopal, Convener, 9442631004
CeBIT INDIA 2016 – Global Event for Digital Business in association with CSI Venue: BIEC, Bengaluru   Contact : Mohammed Farooq,, +91 9004691833
23-24, 2016 8th Annual IEEE International Conference on Computational Intelligence and Communication Network
CICN-2016. Venue : Gyan Ganga Institute of Technology & Sciences, Jabalpur
Contact : Dr. Santosh Vishwakarma
FEBRUARY International conference on Data Engineering and Applications-2017 (IDEA-17) at Bhopal (M.P.),
11-12, 2017  Contact :
MARCH INDIACOM 2017, Organized by Bharati Vidyapeeth’s Institute of Computer Applications and Management
01-03, 2017 (BVICAM), New Delhi
Contact : Prof. M. N. Hoda,,, Tel.: 011-25275055
MAY ICSE 2017 - International Conference on Soft Computing in Engineering, Organized by : JECRC, Jaipur,
08-10, 2017 Contact : Prof. K. S. Raghuwanshi,, Mobile : 9166016670

u 51 u


Registered with Registrar of News Papers for India - RNI 31668/1978 If undelivered return to :
Regd. No. MCN/222/20l5-2017 Samruddhi Venture Park, Unit No.3,
Posting Date: 10 & 11 every month. Posted at Patrika Channel Mumbai-I 4th floor, MIDC, Andheri (E). Mumbai-400 093
Date of Publication: 10th of every month



Amazing speakers and technical

FOR CSI-2016? sessions by world class experts


Hurry! Early Bird registration 3 days of opportunity to personally
closes on 30 Sep 2016 meet and network with industry leaders

Download An extensive range of contemporary
the convention topics will give you a wider view
app CSI-2016
from playstore Bring your family along & take a winter
and iOS & break in Ooty, Munnar or Kodaikanal
stay updated
Get a first hand experience of the world
renowned Kongu region’s warmth

Register online:

For more details