Professional Documents
Culture Documents
CMBEBIH 2017 Proceedings of The International Conference On Medical and Biological Engineering 2017 1st Edition Almir Badnjevic (Eds.)
CMBEBIH 2017 Proceedings of The International Conference On Medical and Biological Engineering 2017 1st Edition Almir Badnjevic (Eds.)
https://textbookfull.com/product/proceedings-of-international-
conference-on-computational-intelligence-and-data-engineering-
iccide-2017-1st-edition-nabendu-chaki/
https://textbookfull.com/product/proceedings-of-5th-
international-conference-on-advanced-manufacturing-engineering-
and-technologies-newtech-2017-1st-edition-vidosav-majstorovic/
https://textbookfull.com/product/proceedings-of-the-2nd-
international-conference-on-data-engineering-and-communication-
technology-icdect-2017-anand-j-kulkarni/
Principles of Security and Trust 6th International
Conference POST 2017 Held as Part of the European Joint
Conferences on Theory and Practice of Software ETAPS
2017 Uppsala Sweden April 22 29 2017 Proceedings 1st
Edition Matteo Maffei
https://textbookfull.com/product/principles-of-security-and-
trust-6th-international-conference-post-2017-held-as-part-of-the-
european-joint-conferences-on-theory-and-practice-of-software-
etaps-2017-uppsala-sweden-april-22-29-2017-pro/
https://textbookfull.com/product/cyber-security-cryptography-and-
machine-learning-first-international-conference-cscml-2017-beer-
sheva-israel-june-29-30-2017-proceedings-1st-edition-shlomi-
dolev/
Volume 62
CMBEBIH 2017
Proceedings of the International Conference
on Medical and Biological Engineering 2017
IFMBE Proceedings
Volume 62
Series editor
James Goh
Deputy Editors
Fatimah Ibrahim
Igor Lacković
Piotr Ładyżyński
Emilio Sacristan Rock
The International Federation for Medical and Biological Engineering, IFMBE, is a federation of national and transnational
organizations representing internationally the interests of medical and biological engineering and sciences. The IFMBE is a
non-profit organization fostering the creation, dissemination and application of medical and biological engineering knowledge
and the management of technology for improved health and quality of life. Its activities include participation in the formulation
of public policy and the dissemination of information through publications and forums. Within the field of medical, clinical,
and biological engineering, IFMBE’s aims are to encourage research and the application of knowledge, and to disseminate
information and promote collaboration. The objectives of the IFMBE are scientific, technological, literary, and educational.
The IFMBE is a WHO accredited NGO covering the full range of biomedical and clinical engineering, healthcare, healthcare
technology and management. It is representing through its 60 member societies some 120.000 professionals involved in the
various issues of improved health and health care delivery.
IFMBE Officers
President: James Goh, Vice-President: Shankhar M. Krishnan
Past President: Ratko Magjarevic
Treasurer: Marc Nyssen, Secretary-General: Kang Ping LIN
http://www.ifmbe.org
CMBEBIH 2017
Proceedings of the International Conference on Medical
and Biological Engineering 2017
123
Editor
Almir Badnjevic
Faculty of Engineering and Information Technologies
International Burch University
Sarajevo
Bosnia and Herzegovina
International Conference on Medical and Biological Engineering in Bosnia and Herzegovina (CMBEBIH2017) is an exciting,
informative and inspiring conference infusing rigorous scientific and engineering practices. CMBEBiH2017 is the largest
Medical and Biological Engineering gathering in South East Europe. The goal of CMBEBIH 2017 is to serve as a platform for
sharing ideas, experiences, best practices and latest advancements in biomedical engineering and all related fields while staying
true to the conference motto: “Pursuing innovation. Shaping the future.”
The conference is a continuation of the extensive work conducted by the Bosnia and Herzegovina Medical and Biological
Engineering Society to encourage and support development of Medical and Biological Engineering in Bosnia and Herzegovina
and Balkan region. The predecessor conference yielded great results with 7 eminent keynote speakers from all around the
world, 120 oral presentations, 50 poster presentations and 3 international workshops.
Medical and Biological Engineering covers all disciplines of the Biomedical and Clinical Engineering spectrum. Research
and development in these areas are impacting the science and technology by advancing fundamental concepts in medicine.
They are helping us better understand human physiology and function at multiple levels by improving tools and techniques for
the detection, prevention and treatment of diseases. The 2017 program consists of 120 accepted papers on new developments
encompassing the conference theme.
More specifically the parallel scientific sessions cover the topics of biomedical signal processing, biomedical imaging and
image processing, biosensors and bioinstrumentation, bio-micro/nano technologies, biomaterials, biomechanics, robotics and
minimally invasive surgery, cardiovascular, respiratory and endocrine systems engineering, neural and rehabilitation engi-
neering, molecular, cellular and tissue engineering, bioinformatics and computational biology, clinical engineering and health
technology assessment, health informatics, e-health and telemedicine, biomedical engineering education, pharmaceutical
engineering, genetic engineering.
International workshops include topics of new perspectives in health technologies regarding diagnostics and therapy,
management systems in the health sector and discussions on how to build a thriving Medical and Biological Engineering
Industry in Bosnia and Herzegovina and other South East European Countries.
Furthermore, the Conference program is highlighted by the seven keynote lecturer sessions. All sessions discuss different
areas of Medical and Biological (biomedical) engineering and are moderated by prominent experts and professionals in the
area.
The session held by Dr. Ernesto Iadanza (Italy), Chair of IFMBE Clinical Engineering Division discuss the Clinical
Engineering. The title of his lecture is “Biomedical and Clinical Engineers: the key to success is a wide knowledge”.
Keynote lecture in the area of processing of biomedical signals is held by Dr. Ervin Sejdic (USA) and focus on discussing
the usage of big data to understand functional changes in swallowing, gait and handwriting.
This keynote lecture entitled “Neuro-prosthetic technologies to restore movement and communication of people with
paralysis” is held by Dr. Tomislav Milekovic (Croatia). Dr. Milekovic is one of the rising stars of Neuro-prosthetics, a field at
the intersection between Engineering, Mathematics and Neuroscience.
More on neural networks, complex algorithms and new advancements is a focus of prof. Dr. Christopher James
(UK) keynote lecture.
Prof. Dr. Nesrin Hasirci (Turkey) will give lecture entitled “Multifunctional Biomaterials and Tissue Engineering Appli-
cations”, while prof. Dr. Vesna Maras (Montenegro) will focus and discuss advancements in Genetic Engineering.
Nanotechnology, one of the Conference themes will be discussed by prof. Dr. Sefik Suzer (Turkey), leading scientists in
surface science and spectroscopy.
We express our gratitude to the Bosnia and Herzegovina leading academic institutions for supporting this project. Particular
thanks are expressed to the kind support and effort of international organizations such as IFMBE, IEEE and ESEM for
endorsing this project, as well as our publisher Springer Nature. Furthermore, there is a limited number of external sponsors for
which we are sincerely thankful.
VI Preface
Finally, a heartfelt thanks to all of you, the participants, for your outstanding contributions. We wish you all the best with
your ongoing research in advancing medical and biological engineering for humanity. We hope that CMBEBIH 2017 creates
opportunities for professional growth and networking with fellow scientist, engineers and other enthusiasts. It is our vision that
this conference creates a lasting memory of Sarajevo and that the networking enables successful and ongoing collaborations
amongst fellow professionals.
We hope that you will appreciate the CMBEBIH 2017 Proceedings as much as we have and that you take pride in this
outstanding project.
Sincerely,
Almir Badnjevic, Phd.
CMBEBIH 2017 Conference Chair
Sarajevo, 16–18 March, 2017
Organization
Organised by
Endorsed by
Technical Co-sponsors
Sponsored by
Platinum Sponsors
Oracle Bosnia and Herzegovina
Računari d.o.o. Banja Luka, Bosnia and Herzegovina
Infostudio doo Sarajevo, Bosnia and Herzegovina
Gold Sponsors
Verlab doo Sarajevo, Bosnia and Herzegovina
Privredna banka dd Sarajevo, Bosnia and Herzegovina
Silver Sponsors
Symphony Sarajevo, Bosnia and Herzegovina
Erkona doo Sarajevo, Bosnia and Herzegovina
Bronze Sponsor
Bosnalijek dd Sarajevo, Bosnia and Herzegovina
Ministarstvo za obrazovanje nauku i mlade Kantona Sarajevo
Sponsors
La Trobe University, Australia
Biosistemi BiH doo Sarajevo, Bosnia and Herzegovina
Phoenix Pharma, Bosnia and Herzegovina
Medic BH, Bosnia and Herzegovina
BBI Centar Sarajevo, Bosnia and Herzegovina
FDS, Bosnia and Herzegovina
Coca Cola, Bosnia and Herzegovina
Klas dd Sarajevo, Bosnia and Herzegovina
BioIRC, Serbia
D-Med Healthcare, Bosnia and Herzegovina
VIII Organization
Conference Chair
Almir Badnjevic, Bosnia and Herzegovina Medical and Biological Engineering Society
Conference Co-chairs
Dusanka Boskovic, University of Sarajevo, Bosnia and Herzegovina and IEEE Region 8 Vice-Chair for Member Activities
Eddie Custovic, La Trobe University Melbourne Australia and Publication Services and Products Board at IEEE
Mario Cifrek, University of Zagreb, Croatia and President of IEEE Croatia EMB Chapter
Program Committee
Abdulhamit Subasi Effat University, Saudi Arabia
Adnan Beganović International Burch University, Bosnia and Herzegovina
Aida Sapcanin University of Sarajevo, Bosnia and Herzegovina
Aljo Mujcic University of Tuzla, Bosnia and Herzegovina
Aleksandar Karač University of Zenica, Bosnia and Herzegovina
Alma Džubur-Kulenović University Clinical Center Sarajevo, Bosnia and Herzegovina
Alma Sofo University Clinical Center Sarajevo, Bosnia and Herzegovina
Amina Kozarić International Burch University Sarajevo, Bosnia and Herzegovina
Anne Humeau-Heurtier University of Angers, France
Antonio Pedotti Politecnico di Milano, Italy
Atif Hodžić University of Bihać, Bosnia and Herzegovina
Božidar Ferek Medtronic, Croatia
Christopher James University of Warwick, United Kingdom
Damir Marjanović International Burch University Sarajevo, Bosnia and Herzegovina
Damijan Miklavčić University of Ljubljana, Slovenia
Davorka Zavrsnik University of Sarajevo, Bosnia and Herzegovina
Dragan Primorac University of Zagreb, Croatia
Dževad Zečić University of Zenica, Bosnia and Herzegovina
Eddie Custovic La Trobe University, Australia
Edin Mujcic University of Bihac, Bosnia and Herzegovina
Edin Pjanić University of Tuzla, Bosnia and Herzegovina
Edina Vranić University of Sarajevo, Bosnia and Herzegovina
Edo Hasanbegović University Clinical Center Sarajevo, Bosnia and Herzegovina
Elma Veljovic University of Sarajevo, Bosnia and Herzegovina
Edhem Hasković University of Sarajevo, Bosnia and Herzegovina
Elmedin Mešić University of Sarajevo, Bosnia and Herzegovina
Enra Suljic-Mehmedika University Clinical Center Sarajevo, Bosnia and Herzegovina
Enisa Omanović Mikličanin University of Sarajevo, Bosnia and Herzegovina
Erna Karalija University of Sarajevo, Bosnia and Herzegovina
Ernesto Iadanza Florence University, Italy
Ervin Sejdić University of Pittsburg, USA
Fahir Bečić University of Sarajevo, Bosnia and Herzegovina
Farid Ljuca University of Tuzla, Bosnia and Herzegovina
Haris Huseinagić University of Tuzla, Bosnia and Herzegovina
Harun Siljak International Burch University, Bosnia and Herzegovina
Hurija Džudžević-Čančar University of Sarajevo, Bosnia and Herzegovina
Herve Liebgot Universitaire de France & Creatis, France
Organization IX
Lejla Gurbeta
Members
Ahmed Osmanović
Alma Jakupović
Amra Mujkanović
Amra Dautbegović
Berina Alić
Branimir Ećimović
Dijana Sejdinović
Fatima Mašić
Fatima Mrkulić
Haris Čampara
Hana Uzunović
Hana Šahinbegović
Iris Kico
Ivana Granzov
Ivana Todorović
Jasna Nuhić
Kristina Rendić
Layla Abd Elilah
Leila Keškić
Lemana Spahić
Nerma Kantić
Sabahudin Ćordić
Sabina Halilović
Zerina Bešić
Zerina Đozić
Snežana Prorok
Contents
MULTISAB project: a web platform based on specialized frameworks for heterogeneous biomedical
time series analysis - an architectural overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Kresimir Friganovic, Alan Jovic, Kresimir Jozic, Davor Kukolja, and Mario Cifrek
Short-term variations of parameters of heart rate variability in subjects with mild hypertension
and normotensive subjects during preoperative period . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Omerbegovic Meldijana
Cardiac pulse waves modeling and analysis in laser Doppler perfusion signals of the skin microcirculation . . . . . . . 20
Michele Sorelli, Antonia Perrella, and Leonardo Bocchi
Discrimination of Psychotic Symptoms from Controls Through Data Mining Methods Based on Emotional
Principle Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Abdullah MARAŞ and Serap AYDIN
Differences in temporal gait parameters between multiple sclerosis and healthy people . . . . . . . . . . . . . . . . . . . . . 31
Kristina Daunoravičienė, Vigita Apanskienė, Jurgita Žižienė, Agnė Ovčinikova, Rasa Kizlaitienė,
Ieva Sereikė, Gintaras Kaubrys, Donatas Lukšys, and Julius Griškevičius
An Adaptive Scheme for X-ray Medical Image Denoising using Artificial Neural Networks and Additive
White Gaussian Noise Level Estimation in SVD Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Emir Turajlić and Vedran Karahodzic
Using Neural Networks and Ensemble Techniques based on Decision Trees for Skin Permeability Prediction. . . . . . 41
Emir Bušatlić, Ahmed Osmanović, Alma Jakupović, Jasna Nuhić, and Adnan Hodžić
Fully Automated Brain Tumor Segmentation and Volume Estimation Based on Symmetry Analysis
in MR Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
C. Öğretmenoğlu Fiçici, O. Eroğul, and Z. Telatar
Multi-Regional Adaptive Image Compression (AIC) for Hip Fractures in Pelvis Radiography . . . . . . . . . . . . . . . . . 61
Huseyin Nasifoglu, Osman Erogul, Gokce Kaan Atac, and Galip Ozdemir
Eustachian Tube Dysfunction Assessment Through Tympanic Cavity Air Exchange Sensor . . . . . . . . . . . . . . . . . . 97
Tarik Ozkul and Murat H. Ozkul
Design, Simulation and Implementation of a Selective Recording Systemfrom Peripheral nervous system . . . . . . . . 104
Hamed Taghipour Farshi and Javad Frounchi
A NOVEL ACTIVE DEVICE FABRICATION METHOD FOR INTERVENTIONAL MRI PROCEDURES. . . . . . 122
Dursun Korel Yildirim, Merdim Sonmez, Robert J. Lederman, and Ozgur Koca-turk
Experimental verification of EOG signal measurement using the modified digital stochastic measurement method. . . 129
J. Djordjević-Kozarov, P. Sovilj, V. Vujičić, D. Mitić, M. Simić, and D. Radenković
Quantification of protein concentration adsorbed on gold nanoparticles using Artificial Neural Network. . . . . . . . . . 142
Adnan Fojnica, Ahmed Osmanović, Dzevida Tarakčija, and Selma Demirović
Design and Fabrication of a PDMS Microfluidic Device for Titration of Biological Solutions. . . . . . . . . . . . . . . . . 147
Medina Dzebic, Olzhas Kurikov, Oleksander Dobroliubov, and Omar C.A. Nava
BEAUTY OF FINE DOTS: Detection and treatment of Alzheimer's disease using quantum dots . . . . . . . . . . . . . . 153
Anida Devedžić
Development of the method for quantification of amino acid adsorbed on nanoparticle surface . . . . . . . . . . . . . . . . 171
Amra Dorić, Džana Mulaomerović, Adnan Fojnica, Belma Alispahić, Sabina Halilović,
and Enisa Omanović-Mikličanin
Contents XIII
Implementation and Validation of Human Kinematics Measured Using IMUs for Musculoskeletal
Simulations by the Evaluation of Joint Reaction Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Maximilian Aurbach, Kilian Wagner, Franz Süß, and Sebastian Dendorfer
FEA of the transiliacal internal fixator as an osteosynthesis of pelvic ring fractures . . . . . . . . . . . . . . . . . . . . . . . . 212
Dominik Jungtäubl, Paul Schmitz, Simon Gross, and Sebastian Dendorfer
Subclinical inflammation: The link between increased cardiovascular risk and subclinical hypothyroidism
in postmenopausal women . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Amina Godinjak, Zelija Velija-Ašimi, Azra Bureković, Mehmed Kulić, Selma Gicić, and Fadila Serdarević
Computational Vascular Surgery Planning and Predicting for Abdominal Aortic Aneurysm . . . . . . . . . . . . . . . . . . 241
Nenad Filipovic, Igor Koncar, Zarko Milosevic, and Lazar Davidovic
Determination of Probabilistic Neural Network’s Accuracy in Context of Cardiac Stress Test . . . . . . . . . . . . . . . . . 246
Sabina Baraković, Jasmina Baraković Husić, and Fahir Baraković
Effects of electrical stimulation as a new method of treating diabetes on animal models: Review . . . . . . . . . . . . . . 253
Hena Divanović, Dženan Mulić, Asima Padalo, Edita Rastoder, Šejla Pedljak, Nermina Žiga, and Tamer Bego
A mathematical model of the effect of metabolic control on joint mobility in young type 1 diabetic subjects . . . . . . 355
Piergiorgio Francia, Antonia Perrella, Michele Sorelli, Sonia Toni, Barbara Piccini, Gabriella Sardina,
Massimo Gulisano, and Leonardo Bocchi
Basics of mathematical modeling of pulmonary ventilation mechanics and gas exchange . . . . . . . . . . . . . . . . . . . . 360
Nataša Popović, Milica Naumović, and Sonja Roganović
The use of ELM and MnM servers for the prediction of RANK function in osteoclast formation . . . . . . . . . . . . . . 372
Sarah Alibegović, Adlija Čaušević, Stefan Kulaš, Tilen Kranjc, Janja Marc, and Tamer Bego
QSAR modeling and structure based virtual screening of new PI3K/mTOR inhibitors
as potential anticancer agents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
Jelena Oluić, Katarina Nikolic, Jelica Vucicevic, Zarko Gagic, Slavica Filipic, and Danica Agbaba
E-health in Bosnia and Herzegovina: exploring the challenges of widespread adoption . . . . . . . . . . . . . . . . . . . . . 388
Borka Popović and Mirjana Maksimović
5-HIAA and HVA in the Coma Cerebri, Hydrocephalus and Tumor Cerebri . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
Salihović M. and Sofić E.
Detremination of kinetic effect of Metoprolol and Ranitidine on HRP- modified GC electrode biosensor . . . . . . . . . 410
Zerina Bešić, Safija Herenda, Trivo Stanković, and Jelena Ostojić
Measuring the feeling: correlations of sensorial to instrumental analyses of cosmetic products . . . . . . . . . . . . . . . . 425
Elezović Alisa, Hadžiabdić Jasmina, Rahić Ognjenka, and Vranić Edina
Hydrophilic antioxidant scores against hydroxyl and peroxyl radicalsin honey samples from Bosnia
and Herzegovina . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
Ismet Tahirović, Dženita Helbet, Adisa Gaštan, Nermin Buza, Muamer Dizdar, Anela Topčagić,
Jasmin Toromanović, Amira Čopra-Janićijević, and Harun Kurtagić
Polymorphisms of 1691G>A and 4070A>G FV in Bosnian women with pregnancy loss . . . . . . . . . . . . . . . . . . . . 435
Mahmutbegovic Emir, Adler Grażyna, Edin Medjedovic, Mahmutbegovic Nevena, Serkan Dogan,
Pawińska-Matecka Anna, Czerska Ewa, and Damir Marjanovic
Lack of association between I/D ACE and -675 ID 4G / 5G PAI-1 polymorphisms and predicting risk
of pregnancy loss (PROPALO) in Bosnian women . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
Mahmutbegovic Emir, Skonieczna-Żydecka Karolina, Valjevac Amina, Mahmutbegovic Nevena,
Pawińska-Matecka Anna, Czerska Ewa, Marjanovic Damir, and Adler Grażyna
Gene clustering using Gene expression data and Self-Organizing Map (SOM) . . . . . . . . . . . . . . . . . . . . . . . . . . . 445
Leila Keškić, Jasin Hodžić, and Belma Alispahić
Public opinion toward GMOs and biotechnology in Bosnia and Herzegovina . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452
Lejla Bevanda, Monika Žilić, Branimir Ećimović, and Vlatka Matković
Future trends and possibilities of using induced pluripotent stem cells (iPSC) inregenerative medicine . . . . . . . . . . . 459
A. Tucak, Dž. Vrabac, A. Smajić, and A. Sažić
XVI Contents
ECG Signal Classification Using Artificial Neural Networks: Comparison of Different Feature Types . . . . . . . . . . . 467
J. Memić
Surface EMG Signal Classification by Using WPD and Ensemble Tree Classifiers . . . . . . . . . . . . . . . . . . . . . . . . 475
Amnah A. Abdullah, Abdulhamit Subasi, and Saeed Mian Qaisar
Tool for Comparative Case Studies of Heart Rate and Heart Rate Variability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482
Haris Ackar, Iris Kico, Anel Tahirbegovic, and Dusanka Boskovic
A novel approach for parameter estimation of Fricke-Morse model using Differential Impedance Analysis . . . . . . . . 487
Mitar Simić, Zdenka Babić, Vladimir Risojević, and Goran Stojanović
Wavelet and Teager Energy Operator (TEO) for Heart Sound Processing and Identification . . . . . . . . . . . . . . . . . . 495
Aiša Ramović, Lejla Bandić, Jasmin Kevrić, Emina Germović, and Abdulhamit Subasi
Ovary Cancer Detection using Decision Tree Classifiers based on Historical Data of Ovary Cancer Patients. . . . . . . 503
Ahmed Osmanović, Layla Abdel-Ilah, Adnan Hodžić, Jasmin Kevric, and Adnan Fojnica
Human-machine interface via EMG signals derived from EEG measurement device . . . . . . . . . . . . . . . . . . . . . . . 516
Nina Maljković, Sofya Pantović, and Jovan Mitrović
A Novel Feature Extraction Approach with VBM 3D ROI Masks on MRI. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523
Muhammet Üsame Öziç, Seral Özşen, and Ahmet Hakan Ekmekci
A modified fuzzy C means algorithm for shading correction in craniofacial CBCT images . . . . . . . . . . . . . . . . . . . 531
Awais Ashfaq and Jonas Adler
Detection and Segmentation of Nodules in Chest Radiographs Based on Lifetime Approach . . . . . . . . . . . . . . . . . 539
Hayati Ture and Temel Kayikcioglu
Automated Colony Counting Based on Histogram Modeling Using Gaussian Mixture Models . . . . . . . . . . . . . . . . 548
Igor S. Maretić and Igor Lacković
Plenary Lectures II - Session IX: CLINICAL ENGINEERING AND HEALTH TECHNOLOGY ASSESSMENT
A testbed evaluation of MAC layer protocols for smart home remote monitoring
of the elderly mobility pattern. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 568
Alireza Jafari, Mohsen Shirali, and Mona Ghassemian
Proposal of integrated software system for simulation and GIS visualization of accidents caused
by emission of hazardous gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 576
Živorad Kovačević and Radovan Stojanović
Contents XVII
Pre-classification process symptom questionnaire based on fuzzy logicfor pulmonary function test cost reduction . . . 608
Almir Badnjevic, Lejla Gurbeta, Mario Cifrek, and Leandro Pecchia
Artificial Neural Network and Docking Study in Design and Synthesis of Xanthenes as Antimicrobial Agents . . . . . 617
Elma Veljović, Selma Špirtović-Halilović, Samija Muratović, Amar Osmanović, Almir Badnjević, Lejla Gurbeta,
Berina Tatlić, Zerina Zorlak, Selma Imamović, Đenana Husić, and Davorka Završnik
Using Information and Communications Technology as an Enabler for Designing an Efficient National
Level Vaccination Planning and Dispensing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 637
Alishba Siab Khan and Shoab Ahmed Khan
Health service quality measurement from patient reviews in Turkish by opinion mining. . . . . . . . . . . . . . . . . . . . . 649
Migena Ceyhan, Zeynep Orhan, and Elton Domnori
e-Medical Test Recommendation System Based on the Analysis of Patients’ Symptoms and Anamneses . . . . . . . . . 654
Migena Ceyhan, Zeynep Orhan, and Elton Domnori
Wireless Body Area Network Studies for Telemedicine Applications Using IEEE 802.15.6 Standard. . . . . . . . . . . . 666
Hasan Yavuz Özderya, Hakan Erdöl, Temel Kayıkçıoğlu, Ali Özgür Yılmaz, and İsmail Kaya
Micro cell culture analog Apparatus (lCCA) output prediction using microcontroller system based
on a Artificial Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 681
Halida Avdihodzic, Sabina Halilovic, Mehrija Hasicic, and Damir Bilk
Multi-biophysical event detection using blind source separated audio signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 690
Jonathan Stanger, Matthew Felicetti, Michael Jenkins, and Edhem Custovic
Diversity performance of microstrip patch antennas placed on human body at ISM and MBAN frequencies. . . . . . . 698
Mehmet Ali Öztürk, Ufuk Tamer, Ali Özgür Yılmaz, Tayfun Özdemir, and İsmail Kaya
AN APPLICATION OF KINECT DEPTH SENSOR FOR SCOLIOSIS AND KYPHOSIS SCREENING . . . . . . . . 713
Octavian Ciobanu, Gabriela Ciobanu, and Antonella Aiello
A Dynamic Stopping Algorithm for P300 Based Brain Computer Interface Systems . . . . . . . . . . . . . . . . . . . . . . . 723
Eda AKMAN AYDIN, Ömer Faruk BAY, and İnan GÜLER
Effects of various metal and drug agents on excretion of enzyme aspartyl proteinase in Candida albicans
and its role in human physiological processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 731
Mirza Ibrišimović, Nadira Ibrišimović-Mehmedinović, Jasmina Dedić, Aldina Kesić, Snježana Marić,
and Almir Šestan
Passive absorption prediction of transdermal drug application with Artificial Neural Network . . . . . . . . . . . . . . . . . 756
Vedad Terzić, Arnela Tarakčija, Almir Vardo, Alma Hadžajlić, Vildana Šakić, Sabrina Smajlović,
Ana-Marija Milisav, Elma Midžić, Alisa Smajović, and Mirza Dedić
The role of population pharmacokinetic analysis in rational antibiotic therapy in neonates . . . . . . . . . . . . . . . . . . . 762
Duspara Kristina, Smits Anne, Allegaert Karel, Heljic Suada, Turcic Petra, Rakovac Tupkovic Lejla,
Kusturica Jasna, Rakanovic-Todic Maida, Burnazovic-Ristic Lejla, Maleskic Sanita, and Kulo Aida
The ratio of hematological parameters and markers of inflammation in patients with iron deficiency
and pernicious anemia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 769
Berina Hasanefendić, Almedina Hajrović, Ermin Begović, and Harun Uzunalić
Contents XIX
Free fatty acid profile in Type 2 diabetic subjects with different control of glycemia . . . . . . . . . . . . . . . . . . . . . . . 781
S. Mandal, A. Causevic, H. Dzudzevic-Cancar, and S. Semiz
Successful collection of stem cells in one day in the process of autologous stem cell transplantation . . . . . . . . . . . . 796
Alma Sofo-Hafizovic, Aida Dizdarevic-Rekic, Lejla Ibricevic-Balic, Lejla Burazerovic, Meliha Vila,
EminaSuljovic-Hadzimesic, Marina Skuric-Tomic, Elma Catovic-Baralija, Gorana Ahmetovic,
Amela Sahovic, Sanela Sicanica, and Amina Omanovic
Abstract. This research implements a feed forward back Early breast cancer detection, is one very important area
propagation network (FFBPN) for classification of breast cancer that researchers are working on, and it can increase the rate
cases to malignant or benign. The purpose of the research is to of diagnosis, treatment and survival of the affected women
design an Artificial Neural Network (ANN) with high and [4]. Considering the high cost of treatment and the high
acceptable level of accuracy by selecting the number of hidden prevalence of the disease among women worldwide, early
layers, number of neurons in the hidden layer and the type of
diagnosis is found to be the most effective step in reducing
activation functions in hidden layers. Samples for training and
validation of ANN are obtained from Wisconsin Breast Cancer
the health and social complications of this disease. There
Database (WBCD) which is open access dataset. The dataset are many different methods and procedures to detect this
contains 699 samples that were distributed to two groups: 599 type of cancer, with different advantages but also disad-
samples in training setand 100 samples in testing set. Each vantages. Cancer is usually detected and diagnosed when
sample has 9 attributes representing 9 characteristics of breast fine- the cancer has progressed to its later stages. This is particu-
needle aspirates (FNAs) as inputs of the network. This experiment larly bad because upon discovery, the chances that cancer
includes a comparison among the obtained mean square error has metastasized are high making the chances of curing it
(MSE) when using three transfer functions: LOGSIG, TANSIG, very low. People rarely perform self-tests which leads to
and PURELINE in neural network architetcures.Impact of detecting cancer in its later stages. Most breast cancers are
different number of layers (1, 2, and 3 layers were used)in ANN detected as a lump/mass on the breast, or through self-
architectureon output accuracy was also investigated. Also, this
examination or mammography [5].
research provides the results of ANN performance for different
number of neurons in hidden layer (20, 21, 22, 23, 24 neurons
Automated systems that can assist professionals in de-
were implemented). The results show that the best network design tecting certain disease can help in early diagnosis and can
is that one with three hidden layers, 21 neurons in the hidden layer, facilitate the treatment process. Developing tools for disease
and TANSIG as activation function. classification can be extensive and challenging, especially
when the association between diagnosis parameters and
Keywords: Artificial Neural Network, breast cancer, disease classification is non-linear and de-pends on various
classification, machine learning, malignant, benign, accuracy, fine- factors. The usage of artificial intelligence techniques in
needle aspirates, Wisconsin Breast Cancer Database (WBCD). disease classification have been in focus of research exten-
sively in the past few years. Artificial Neural Networks
(ANN) can be used for this purpose. The idea behind these
1 Introduction systems lies in parallel information analyzing in structures
that mimic biological nervous system. The main units, neu-
Cancer, also known as malignant neoplasm, is a class of rons, can be connected in feed forward and feedback archi-
over a hundred diseases characterized by abnormal growth tectures [6]. The activation of each neuron is determined by
of cells with a potential to spread to surrounding tissues of linear or non-linear transfer function. ANNs have the ability
the body [1]. It is a vicious disease which is responsible for to perform many functions: classification, fault detection,
numerous deaths in both developing and developed coun- speech analysis, and processing of inaccurate inputs [7].
tries of the world. Breast cancer is the second leading cause Over the past few years, these methods have been used,
of cancer death among females in more developed coun- more extensively, in disease classification to improve the
tries, and it is the leading cause of cancer death among fe- diagnoses procedures and to aid the physician’s efforts [8].
males in less developed countries. [2]. some new statistics For example, ANNs were used in [9] in neurodegenerative
found that about 1 in 8 women in the US develop breast disease classification, cardiac disorders [10], predictions on
cancer during their lifetime [3]. Thus, the fight against can- a Nano-scale [11] as well as different respiratory disease
cer is far from completed. classification [12-14].
© Springer Nature Singapore Pte Ltd. 2017 3
A. Badnjevic (ed.), CMBEBIH 2017,
IFMBE Proceedings 62,
DOI: 10.1007/978-981-10-4166-2_1
4 L. Abdel-Ilah, H. Šahinbegoviü
2.1 Wisconsin Breast Cancer Database (WBCD) Artificial Neural Networks (ANNs) are a strong tool for
classificationof cancer [16-18] and are very clear examples
This breast cancer database, used in this research, is col- of using ANNs in classification purposes for different types
lected from the UCI machine-learning repository, which of cancer: liver, colon, and skin cancers respectively with a
was developed by Dr. William H. Wolberg from the Uni- successful results.When designing neural network for solv-
versity of Wisconsin Hospitals, Madison [15]. The dataset is ing specific problem, network architecture, number of lay-
comprised of elements that have various scalar observa- ers, number of neurons in hidden layers, training dataset
tions. The total number of the samples is 699. The dataset distribution and training algorithm have significant impact
contains two classes referring to benign and malignant sam- on system performance [19]. A feed forward neural network
ples. There are 458 samples in the dataset that are assigned is most commonly used for classification and prediction
to benign and the other 241 samples are malignant, the total purposes [20].
number of samples is used in this research. The original In this study, not only impact of number of neurons in
dataset contains 11 attributes including both sample ID hidden layer on system performance is examined, but also
number and class label, which are deleted in the dataset that the impact of hidden layer number and type of activation
are used in this work since it can just cause disruption. The function in neuron. Based on previously mentioned recom-
remaining 9 attributes represent 9 characteristics of breast mendations from experts, in this study, feed forward archi-
fine-needle aspirates (FNAs), as shown in Table 1. tecture is chosen for breast cancer classification. Number of
A fine-needle aspirates (FNA) is performed for the diag- input and output neurons is determined with parameters that
nosis of lesions, which requires the removal of cells from a are used for disease classification. Trial method with chang-
breast lesion or cyst with a fine needle. The cytological ing number of neurons and comparing performance between
characteristics of breast FNAs were valued on a scale of one neural network architecture is used. Number of neurons in
to ten, with one being the most benign and ten the closest to hidden layer is set to be 20, 21, 22, 23 and 24 and perfor-
malignant. These attributes measure the physical appearance mance in terms of Mean Square Error (MSE) and Absolute
and chromosome changes in the cell. There are two values Error was compared.
in the class variable of breast cancer: Benign (non- ͳ
ଶ
cancerous), and Malignant (cancerous). ܧܵܯൌ ൫ܺௗ௧ௗ െ ܺ௧௨ ൯
݊
ୀଵ
mance criteria, but when the dataset is limited, division ratio 3.2 Predictive ability of neural network architecture
may play a crucial role. For smaller database most research- with different number of neurons in hidden layer
ers suggest 70-30 distribution [21]. In this study, the
devision is about 85-15. Performance in terms of accuracy in breast cancer classi-
For neural network training, Levenberg- Marquardt fication for neural network architecture with one hidden
algorithm was used which approves a high performance and layer and 20, 21, 22, 23, 24 hidden neurons with all other
minimized error in disease classification [22]. At the start of parameters fixed is presented in Table 3. As it can be seen
training, all connection weights in network are set to from Table 3. The network with 21 neurons achieves the
random values. All input vectors are normalized so that the highest accuracy (98%), whereas the one with 22 neurons
minimum and maximum are 0 and 1 respectively. has the lowest accuracy (95%).
Normalization was done according to:
Table 4. Second group contains five networks with different
ܰ݁ݑ݈ܽݒ݀݁ݖ݈݅ܽ݉ݎ number of neurons and different accuracies of testing outputs.
ܴ݈݁ܽ ݁ݑ݈ܽݒെ ݐ݁ݏܽݐܽ݀݊݅݁ݑ݈ܽݒ݉ݑ݉݅݊݅ܯ
ൌ Index Number of Number of Number of Accuracy
ݐ݁ݏܽݐܽ݀݊݅݁ݑ݈ܽݒ݉ݑ݉݅ݔܽܯെ ݐ݁ݏܽݐܽܦ݊݅݁ݑ݈ܽݒ݉ݑ݉݅݊݅ܯ
neurons correct incorrect %
results results
1. 20 97 3 97
3 Results and discussion
2. 21 98 2 98
3. 22 95 5 95
3.1 Predictive ability of neural network architecture 4. 23 96 4 96
with different number of hidden layers 5. 24 96 4 96
Fig. 1. MSE
M results afteer training the networks
n impleementing a) TA
ANSIG, b) LOG
GSIG, c) PUR
RELINE activattion
ffunctions. The MSE values arre: 0.000140366, 0.0071772, and
a 0.0089316 respectively. TANSIG
T functtion achieves thhe best
perfformance.
During the designing proceess of an acccurate artificiaal parameterrs will providde us with moore accurate network,
n
neuural network forf breast canncer classificattion, it is veryy which will help in claassifying breaast cancer casees more
im
mportant to takee care about thee different paraameters. As wee efficiently
y and achievinng medical, soocial, and ecoonomical
cann notice, all neetworks in thiss experiment have
h a feed for- benefits.
waard back propaagation architeecture but impplementing diff-
ferrent number off hidden layerss, neurons in thhe hidden layerr
andd different typpes of activatioon functions leead to differennt 4 Conclusion
perrformance resuults. High perfformance cannnot be achievedd
by focusing on one o parameter only like in [24], where thee Even though
t there iss no generalizeed neural netwoork con-
perrformance of the network was w comparedd among threee figuration
n with zero mistakes, for a particular appliccation, a
nettworks with diifferent architeectures: Multi Layer Percep- neural network with a high
h and accepttable level of accuracy
a
troon (MLP), Genneralized Regreession (GRNN)) and Probabil- can be designed
d by manipulating
m thhe number off hidden
istiic (PNN) archhitectures. Ressults state thatt that the most umber of neuroons in the hiddden layer and the type
layers, nu
suiitable neural network
n modell for classifyinng WBCD aree of activattion functions. This experimeent includes a compari-
c
ML LP and GRNN N, and observinng that PNN peerformance wass son amon ng the obtaineed MSE as a result of trainning net-
loww in case of smmall datasets without
w mentiooning the effect works applying
a threee transfer functions: LOGSIG,
L
of the other paraameters. In [20], it is also a comparison
c de- TANSIG,, and PUREL LINE. Also, inncludes a com mparison
pennding on the type of archittecture only, and a the resultss among teesting set accurracies obtainedd by using 1, 2, 2 and 3
staate that the baack propagatioon achieves thhe best perfor- layer/s neetworks. Last comparison
c is among testingg set ac-
maance without coomparing otherr parameters. Improving thee curacies obtained
o from networks conttain 20, 21, 222, 23, 24
nettwork perform mance by foccusing on moore and moree neurons.
Results show that best network design is that one with power systems. In Power Electronics, Drives and Energy Sys-
three hidden layers and 21 neurons in the hidden layer. In- tems, 2006. PEDES'06. International Conference on (pp. 1-9).
creasing hidden layers number or neurons number does not IEEE.
mean for necessary better accuracy. From the three activa- 6. Techopedia, Artificial neural network, https://www.
techopedia.com/definition/5967/artificial-neural-network-ann.
tion functions LOGSIG, PURELIN AND TANSIG provid-
Accessed: 4 Feb 2016
ed, the network with TANSIG has produced the least MSE
7. Singh, S., & Murthy, T. V. (2013). Neural network-based sen-
i.e. 0.0001 which is very low error as compared to other sor fault accommodation in flight control system. Journal of
functions i.e. TANSIG function is most suitable activation Intelligent Systems, 22(3), 317-333.
function for WBCD dataset. It has been noticed that as the 8. Hiyama, T. (1990). Rule-based stabilizer for multi-machine
number of times we train our network model MSE decreas- power system. IEEE Transactions on Power Systems, 5(2),
es, but overtraining may also lead to increase in error. 403-411.
In future there is a possibility to study different parame- 9. Aljovic, A., Badnjevic, A., Gurbeta, L. (2016, June). Artificial
ters and find their effect on the results for WBCD data set. neural networks in the discrimination of Alzheimer's disease
By studying all the aspects we can design a better algorithm using biomarkers data. In Embedded Computing (MECO),
and tool for neural network. Also developing GUI (graph- 2016 5th Mediterranean Conference on (pp. 286-289). IEEE.
10. Gurbeta, L., Sajdinovic, D., Berina, A., & Badnjevic, A.
ical User interface) can be a strong step toward creating
(2016, June). Classification of stress recognition using Artifi-
proper automated system for breast cancer classification and
cial Neural Network. In Embedded Computing (MECO),
validation of this system on real time with acquired data 2016 5th Mediterranean Conference on (pp. 297-300). IEEE.
provided by hospitals in Bosnia and Herzegovina. 11. Haliloviü, S., Avdihodžiü, H., Gurbeta, L. (2016, June). Micro
Designing a better algorithm will provide us a possibility cell culture analog apparatus (ýCCA) output prediction using
of replacing some old and less efficient classification tech- Artificial Neural Network. In Embedded Computing (MECO),
niques by ANN which will help in time, cost, and life sav- 2016 5th Mediterranean Conference on (pp. 294-296). IEEE.
ing by providing the patients with early diagnosis, and re- 12. Fojnica, A., Osmanoviü, A., & Badnjeviü, A. (2016, June).
duce the pressure on clinics and hospitals. Moreover, it will Dynamical Model of Tuberculosis-Multiple Strain Prediction
reduce the economic and mental burden on patients as a based on Artificial Neural Network. In Embedded Computing
result of a very long treatment of breast cancer. (MECO), 2016 5th Mediterranean Conference on (pp. 290 -
293). IEEE.
13. Badnjevic, A, Cifrek, M., Koruga, D., Osmankovic, D. (2015)
Conflict of interest Neuro-fuzzy classification of asthma and chronic obstructive
pulmonary disease. BMC Medical Informatics and Decision
Making Journal. 15 Suppl 3:S1; doi: 10.1186/1472-6947-15-
The authors declare that they have no conflict of interest.
S2-S1
14. Badnjevic, A., Cifrek, M., & Koruga, D. (2014). Classifica-
tion of Chronic Obstructive Pulmonary Disease (COPD) using
References integrated software suite. In XIII Mediterranean Conference
on Medical and Biological Engineering and Computing
1. Klonisch, T., Wiechec, E., Hombach-Klonisch, S., Ande, S. 2013 (pp. 911-914). Springer International Publishing.
R., Wesselborg, S., Schulze-Osthoff, K., & Los, M. (2008). 15. UCI Machine Learning Repository, Breast Cancer Wisconsin
Cancer stem cell markers in common cancers–therapeutic im- (Original) Data Set,’’https://archive.ics.uci.edu/ml/datasets/
plications. Trends in molecular medicine, 14(10), 450-460. Breast+Cancer+Wisconsin+(Original). Accessed: 4 Feb 2016
2. Torre, L. A., Bray, F., Siegel, R. L., Ferlay, J., Lortet- 16. Hafizah, S., Haszlinna, N., Mustaffa, Sallehuddin, R., &
Tieulent, J., & Jemal, A. (2015). Global cancer statistics, Ubaidillah, S.A. (2014). Classification of Liver Cancer Using
2012. CA: a cancer journal for clinicians, 65(2), 87-108 Artificial Neural Network and Support Vector Machine.
3. Chandrasekar, R. M., & Palaniammal, V. (2013). Perfor- 17. Ahmed, F. E. (2005). Artificial neural networks for diagnosis
mance and Evaluation of Data Mining Techniques in Cancer and survival prediction in colon cancer. Molecular can-
Diagnosis. IOSR Journal of Computer Engineering, 15(5), 39- cer, 4(1), 29.
44. doi:10.9790/0661-1553944. 18. B, A.R., Jaleel, J.A., & Salim, S. (2012). Artificial Neural
4. Cohen, M., & Azaiza, F. (2005). Early breast cancer detection Network Based Detection of Skin Cancer
practices, health beliefs, and cancer worries in Jewish and 19. Devika, C., Amita, S. M., (2014) Case Study on Classification
Arab women. Preventive medicine, 41(5), 852-858. of Glass using Neural Network Tool in MATLAB.
5. Etingov, P. V., & Voropai, N. I. (2006, December). Applica- International Journal of Computer Applications® (IJCA)
tion of fuzzy logic PSS to enhance transient stability in large (0975 – 8887), International Conference on Advances in
Computer Engineering & Applications (ICACEA) at 23. Rani, K. U. (2010). Parallel approach for diagnosis of breast
IMSEC,GZB cancer using neural network technique. International Journal
20. Swathi, S., Rizwana, S., Babu, G. A., Kumar, P. S., & Sarma, of Computer Applications, 10(3), 1-5.
P. V. G. K. (2012). Classification of Neural Network Struc- 24. Elgader, H. A. A., & Hamza, M. H. (2011). Breast Cancer Di-
tures For Breast Cancer Diagnosis. International Journal of agnosis Using Artificial Intelligence Neural Networks. J. Sc.
Computer Science and Communication, 3(1), 227-231. Tech, 12, 159-171.
21. Shahin, M. A., Maier, H. R., & Jaksa, M. B. (2004). Data di-
vision for developing neural networks applied to geotechnical Corresponding author:
engineering. Journal of Computing in Civil Engineer- Author: Layla Abdel-Ilah
ing, 18(2), 105-114. Institute: International Burch University
22. Khan, N., Gaurav, D., & Kandl, T. (2013). Performance eval- Street: Francuske Revolucije BB
uation of Levenberg-Marquardt technique in error reduction City: Sarajevo
for diabetes condition classification. Procedia Computer Sci- Country: Bosnia and Herzegovina
ence, 18, 2629-2637. Email: Layla.abdelilah@hotmail.com
Kresimir Friganovic*, Alan Jovic*c, Kresimir Jozic**, Davor Kukolja* and Mario Cifrek*
*University of Zagreb Faculty of Electrical Engineering and Computing, Unska 3, 10000 Zagreb, Croatia
**INA – industrija nafte, d.d., Avenija Veceslava Holjevca 10, p.p. 555, HR – 10002 Zagreb, Croatia
kresimir.friganovic@fer.hr, (corr. author): alan.jovic@fer.hr, kjozic@gmail.com, davor.kukolja@fer.hr, mario.cifrek@fer.hr
c
Abstract. The aim of this work is to present an architectural over- used technologies and use case based scenarios specifica-
view of a novel web platform used for heterogeneous biomedical tion [1,2]. The platform predicts a total of 8 fundamental
time series analysis. Its architecture is based on three subprojects: use case scenarios: analysis type selection, scenario selec-
frontend, backend, and processing. Frontend uses several contem- tion, input data selection, records inspection, records pre-
porary web technologies to present a fast, responsive and pleasing
processing, feature extraction, model construction, and re-
user interface. Backend, written in Java, communicates with a
porting [2]. Additional scenarios for handling the user ac-
database and with other servers, on which the processing subpro-
ject is deployed. The processing subproject contains several count and platform administration were also provided. The
frameworks intended for: record input handling, signal prepro- approach that we followed allows adaptive control of user
cessing, data visualization, general time series features extraction, needs, and can be specialized for different kinds of analysis
specific time series features extraction (e.g. heart rate variability goal (e.g. classification, visualization, prediction). As the
and electroencephalogram), data mining, and reporting. The plat- system is under development as a web platform, the users
form is in an early phase of implementation, but we demonstrate will be able to access it from afar and transmit biomedical
its features and capabilities, of which feature extraction frame- data to the server, where efficient analysis can be per-
works and signal visualization currently stand out. formed. Also, a web platform enables easier maintenance
and larger user base. This distinguishes our platform from
Keywords: framework, web platform, biomedical time series, local or hospital-specific software solutions [2].
frontend development, signal visualization The aim of the current paper is to provide an architectural
overview and detailed description of the MULTISAB pro-
ject's frameworks, and to show the currently implemented
1 Introduction visualization aspects of the platform.
The architectural overview of the platform is shown in
The need for biomedical software is showing growth in Fig. 1. MULTISAB project is divided into three subpro-
healthcare community and in general population users. De- jects: frontend, backend and processing. The frontend and
spite a great number of web and mobile applications target- backend subprojects support communication between end
ing general users, such as sports and fitness enthusiasts, users and servers. Basic functions are currently supported:
specialized software developed for medical professionals is web browser login into the system, uploading and visualiza-
limited to continuous monitoring of biomedical time series tion of the heterogeneous BTS, specification of the steps
(BTS). To overcome this problem, we pursue the develop- included in the analysis and defining the outcomes of the
ment of an efficient and upgradeable BTS analysis system analysis. The processing subproject contains several
in a form of an integrative web platform for automatic clas- frameworks for heterogeneous BTS data analysis and will
sification of human body disorders. We call it "MULTISAB be discussed in more details in Section 2. In ideal condi-
project" (short for: Multiple Time Series Analysis in Bio- tions, every subproject runs on different computer host(s),
medicine). The system will serve as a help for medical spe- therefore, they do not share physical computer resources.
cialists in diagnostics and early detection of various diseases TypeScript, which is used on frontend, is a strict superset of
[1]. JavaScript that compiles to the selected version of JavaS-
In our previous work, we described the early efforts in cript. Through his web browser, a user (client) can send
designing such a system, which included an overview of requests to the server implemented on backend.
© Springer Nature Singapore Pte Ltd. 2017 9
A. Badnjevic (ed.), CMBEBIH 2017,
IFMBE Proceedings 62,
DOI: 10.1007/978-981-10-4166-2_2
10 K. Friganovic et al.
The backend subproject is written in Java 8 and serves packages specialized for different tasks in signal analysis,
for processing requests from users and for communication which correspond to the mentioned use case scenarios.
with a database and data analysis hosts. Java 8 is used, be- Detailed description of some of the implemented frame-
cause it enables efficient parallel processing with data works is given in the following subsections. Frameworks
streams. A JPA framework (EclipseLink) is used to estab- for ECG analysis and data mining algorithms currently un-
lish communication between the Java-written backend and a der construction are discussed in Section 5.
relational database by using object-relational mapping. It
also implements a persistent control of data. In order to run
2.1 Common signal features framework
all the necessary resources on the backend host, Spring Boot
framework is used, which facilitates web application im- The framework for common signal features extraction
plementation. In the future, 3D visualization of body disor- includes a great number of features for time series analysis
ders using WebGL on the client-hand side will be support- in time (both linear and nonlinear), frequency and time-
ed. In order to communicate with the graphical processing frequency domain. The EEGframe and HRVframe [3,4]
units in Java, OpenGL/OpenCL are used with jocl Java frameworks have been refactored to extrude general BTS
framework. features for MULTISAB processing into a common signal
Communication between backend and frontend uses features framework. The idea behind such a great number of
HTTPS protocol to insure secure transfer and communica- common features is a more accurate and more all-
tion. For communication between Java 8 on backend and encompassing analysis than it is usually pursued for BTS
JavaScript on frontend, RESTful protocol is used. (comparable to the work of Fulcher et al. [5]). The common
features can be used alone or together with specialized fea-
tures to train data mining classifiers in order to efficiently
2 Processing frameworks classify diseases.
Power spectral density (PSD) of a signal, which is used
The processing subproject is written entirely in Java 8 for frequency domain analysis, is implemented using three
and implements everything regarding data processing and different algorithms: fast Fourier transform (FFT), Burg
analysis. The subproject is divided into several frameworks method and Lomb-Scargle method. Three window types are
in order to improve maintainability and scalability of the supported for FFT-based PSD estimate: rectangular, Hann,
integrated system. The frameworks consist of one or more and Hamming window. Other included transformations are