TITLE

CLASSIFICATION AND DETECTION OF BONE

CANCER IN MEDICAL IMAGES USING NEURAL
NETWORK

ABSTRACT

Automatic defects detection in Mrand CT images is very important in many diagnostic and therapeutic
applications. Because of high quantity data in MR images and blurred boundaries, tumour
segmentation and classification is very hard. This work has introduced one automatic bone tumour
detection method to increase the accuracy and yield and decrease the diagnosis time. The goal is
classifying the tissues to three classes of normal, begin and malignant. . In MR and CT images, the
amount of data is too much for manual interpretation and analysis. During past few years, bone cancer
segmentation in magnetic resonance imaging (MRI) has become an emergent research area in the field
of medical imaging system. Accurate detection of size and location of bone cancer plays a vital role in
the diagnosis of cancer. The diagnosis method consists of four stages, pre-processing of MR images,
feature extraction, and classification. After pre-processing of image, the features are extracted based
onwavelet transformation (DWT). In the lastt Neural Network (NN) are employed to classify the
Normal and abnormal bone. An efficient algorithm is proposed for cancer detection based on the
Spatial Fuzzy C-Means Clustering.

AIM

Detecting and clustering the various neoplasm in our body with the algorithms of
image processing.

OBJECTIVES:

1. Literature survey on cancers part by image processing for the past years.
2. Study about CT and MRI images of medical applications
 3. Study about neural network and its mathematical formulation
4. The installation and simulating of matrix laboratory.
5. Comparison of accuracy with experimental results.

Existing system:

For do this project in existing system we have threshold process to detect the cancer part. The
process is we are taking the input image as MRI and CT images of symptoms of neoplasm . After that
according to our requirement we are interchanging the pixel values, means converting and resizing the
image. By using some filters we are reducing the noise. Here detecting the cancer part is more difficult
to do this we are using K-NN algorithm. We are using the threshold process to clear the structure of the
input image. Afterwards we need to cluster the part to detect what type of cancer it was. Here we are
not using any particular algorithm for clustering manually we are analyzing the process. But it is very
difficult to do the process.

The drawbacks of existing system we have. It is difficult to get the accurate resultsBy using
these techniques. If we applied multiple images at a time, we cant get the results very soon. It
consumption more time to give the results. In medical images by default it self we have some noise, it
is very difficult to filter and classify. It may give inaccuracy classification.

PROPOSED Methodologies

* DWT

* GLCM Feature Extraction

* NN Training and Classification

* SFCM

PROPOSED BLOCK DIAGRAM

Proposedmethodology:

a)Input image

The input image is in DICOM format this image can be convert into JPEG format and resize the image,
because the image is having more size, it requires more time for segmentation process and less picture
quality. So the size should be resized into 256*256. The input images for this work using Bone (MRI)
and Lung (CT) images received from diagnosis hospitals.
b) Preprocessing

Preprocessing is used to improve the quality of an image. Every image has contained some salt and
pepper noise having some blurriness. To remove the noise and blurriness’ using Median filter.

c)Median filter

The median filter is a sliding window spatial filter, it replaces the center value of an window with the
median of all the pixel .Due to changing the median of all the values through the center value it remove
the noise and preserve the edges of an image, its one type of smoothening technique. It improves the
quality of an image. There is no reduction in contrast, it doesn’t shift boundaries and unrealistic values
are not created near edges.

d) Segmentation process

DWT used to detect the cancer in bone (MRI) and lung (CT) images. The both algorithms is used to
segment the cancer from bone and lung images. The image can be segmented thoroughly and finally
obtained the image into segments.

e) Classification

Normally the classification is used to classify that the image is normal or abnormal. NN is one type of
classifier, the features and values of the cancer affected image and non cancer image is already placed
in database, the intensity is also having in cancer affected image, the classifier compares the given
image within the database if the cancer is identified while comparing the each pixels, it display the
message box the cancer is affected, after completing the NN training.

f) Feature extraction

The feature extraction is a major process in recognition applications and classifications, the texture
based feature extraction is going on in this work, normally several texture based feature extraction
classifications are there those are GLCM, LBP,SLBP... The gray scale invariant texture is measured
and derived from definition of texture in local region. It is an efficient texture operator , it labels image
pixels by the threshold process from the neighborhood of each pixel and represents in binary number.
In this the cancer part is extracted from the lung and bone images, this is based on the texture and
contrast of an image. Input cancer image (MRI/CT) Preprocessing Segmentation Using Classification
with KNN Feature extraction Statistical values
Here we are confirming that the stage of the cancer regarding the cancer part. By clustering the cancer
part we can classify and detect the stage of the caner, whether it is in the malignant stage or benign
stage. Here we can consider that benign is the starting stage of the caner we can by detecting this we
can reduce the affection of cancer. But malignant is the final stage of the caner we cant reduce the
condition.

f) Statistical values

The statistical values are determined for both bone and lung images using PSO and CSO techniques.
The MSE, PSNR values are calculated and the MSE is less than the PSNR value. The other parameters
are sensitivity, specificity, accuracy and processing time.

Advantages

* It can segment the Bone regions from the image accurately.

* It is useful to classify the Bone Cancer images for accurate detection.

* Bone Cancer will be detected in an early stages

FUTURE SCOPE

* Video analysis

* Drug discovery
REFERENCES:

[1] John E Hall, Guyton and Hall textbook of medical physiology e-Book, Elsevier Health Sciences,
2015.

[2] Multiple Myeloma Research Foundation (MMRF), “Multiple myeloma research foundation,” 20–.
[3] Jean-Luc Harousseau Rgis Bataille, “Multiple myeloma,” The New England Journal of Medicine,
Massachusetts Medical Society, Jun 5, 1997.

[4] Samer Z Al-Quran, Lijun Yang, James M Magill, Raul C Braylan, and Vonda K Douglas-Nikitin,
“Assessment of bone marrow plasma cell infiltrates in multiple myeloma: the added value of cd138
immunohistochemistry,” Human pathology, vol. 38, no. 12, pp. 1779– 1787, 2007.

[5] Alastair Smith, Finn Wisloff, Diana Samson, Nordic Myeloma Study Group UK Myeloma Forum, and
British Committee for Standards in Haematology, “Guidelines on the diagnosis and management of
multiple myeloma 2005,” British journal of haematology, vol. 132, no. 4, pp. 410–451, 2006.

[6] Omid Sarrafzadeh, Hossein Rabbani, Ardeshir Talebi, and Hossein Usefi Banaem, “Selection of the
best features for leukocytes classification in blood smear microscopic images,” in Medical Imaging
2014: Digital Pathology. International Society for Optics and Photonics, 2014, vol. 9041, p. 90410P.