Undergraduate Thesis

Automatic Diagnosis of Heart Valve

A Thesis Submitted to the Department of Electrical and Electronic Engineering in Partial Fulfillment of the Requirement for the Degree of

Bachelor of Science in Electrical and Electronic Engineering (EEE)

Department of Electrical and Electronic Engineering

United International University, Dhaka, Bangladesh

2011

A thesis on

Automatic Diagnosis of Heart Valve

Key words: Phonocardiogram, Valvular Diseases, Template Based Feature, Neural Network.

By

Al Jumlat Ahmed                          ID# 021063002

Obaidul Haque                              ID# 021063017

Mohammad Fattah Halim         ID# 021063040

Sadia Anjum Jeba                         ID# 021061049

Supervisor

Dr. M.K Masukur Rahman

Assistant Professor,

Department of EEE

United International University   

Dedication

 To our respected Parents & Teachers………..

Declaration

It is hereby declared that this thesis or any part of it has not been submitted elsewhere for the award of any degree or diploma.

Signature of the Supervisor:

                                                                            Dr. M.K Masukur Rahman

Signature of the Candidates:

   Al – Jumlat Ahmed

              Obaidul Haque

   Mohammad Fattah Halim

               Sadia Anjum Jeba

Acknowledgements

We are paying our gratitude to the Almighty Allah because of whom we are being able to prepare this Thesis Report. We earnestly pay our regard to the Almighty to put us into this success.

We would like to express our respect and heartiest gratitude to Dr. M.K. Masukur Rahaman, Asst. Professor, Department of EEE, United International University who is the Supervisor of  our Thesis (EEE - 400),  for his great cooperation, which will be always remembered by us as without whom preparing this report was simply impossible. There are many people who have direct support and contribution to this work. We would like to thank all of them.

Ø Al Jumlat Ahmed

Ø Obaidul Haque

Ø Mohammad Fattah Halim

Ø Sadia Anjum Jeba

Abstract

Heart is the most important organ in human body. Detection of heart valvular diseases by analyzing phonocardiogram signal is simple and cost effective compare to other diagnosis methods. In our work we have developed novel template based features in both time and frequency domain.   Time and Frequency domain template features have been constructed by analyzing the audio signals of heart. To analysis the audio signal we have implemented our own algorithm. Five different conditions of heart valves have been detected successfully by our method and it would be applicable for other heart valve conditions. Multi Layer Perception (MLP) type Neural Network (NN) has been employed to classify different valvular diseases and different valvular diseases have been classified successfully by this classifier. We have used 83 data for performance evaluation of our method among them 50% has been used as training data and rest 50% as testing data. We have evaluated performance of our method using time, frequency and hybrid features and outcome is very satisfactory. Finally we have developed a low cost, simple and automatic diagnosis system of heart valve which required a computer, simple stethoscope and a microphone.