5817

5432

كارشناسي ارشد

مخابرات

اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر

1389

نه،92ص.: مصور،جدول،نمودار

ص.ع.به فارسي و انگليسي

رسول امير فتاحي، بهزاد نظري

مجتبي منصوري

5/2/90

سعيد صدري، محمدرضا احمدزاده

مهندسي برق و كامپيوتر

ID5432

به فارسي و انگليسي: قابل رويت در نسخه ديجيتالي

92 Non invasive Detection of Cardiac output Using Digital Processing of Heart Sounds Mahdi Kheirolahi m kheirolahi@ec iut ac ir Date of Submission 2011 03 15 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language Farsi Supervisor Rassoul Amirfattahi fattahi@cc iut ac ir Behzad Nazari nazari@cc iut ac ir Abstract Listening to the heart sound or in fact hearing sounds which are produced by the heart is a basic tool for heart diseases diagnose This method can provide doctors with worthy information about the function of the heart Diagnosing signs related to the heart diseases and classifying them by listening to the heart sound is a complex trait for doctors even years of practice and experience are necessary for a doctor to make his ear familiar to such sounds and gain necessary skills Recent research indicated that PCG signal has crucial information about time and frequency By correctly analyzing it an appropriate diagnoses can be made for the patient 1 The history of different heart diseases diagnose by using PCG signal dates back to 1983 when professor Simloo and his colleagues used phonocardiography for non invasive diagnose of coronary artery occlusions Cardiac output heart ejection fraction is the amount of blood that per minute is pumped into the vessels body by the heart Cardiac output is multiplied ejection fraction in heart rate and is a good indicator of cardiovascular system conditions that plays a key role in control of cardiovascular diseases and determine the cause hemodynamic impairment quickly and providing appropriate treatment Several methods are presented for measuring cardiac output that some of these methods are invasive such as thermodilution method that is the most common measurement method in clinic Today is trying to diagnostic measures are non invasive treatment possible To this end methods such as Echocardiography Cardio Q Picco Lidco Nicco Bio Z Pulse contour and Hemosonic are presented for measuring cardiac output In this research we propose a non invasive method for assessment of ejection fraction index through digital processing of heart sounds For this reason several features have been extracted from phonocardiograms and then classified In the first step we propose two different algorithms for separation of main sounds of heart called s1 and s2 In the first algorithm heart sounds have been separated using wavelet transform and Shannon energy envelope independent of electrocardiogram signal while in the second QRS complex has been used as a reference for heart sound separation Results show that the specificity of the first algorithm is 100 and 89 3 for normal and abnormal cases respectively The specificity of the second algorithm is 99 87 and 93 44 for normal and abnormal cases respectively in the second step ejection fraction index is evaluated by extraction of 5 features from heart sound signal The extracted features were classified using RBF neural network Results show that the neural network output is matched with the physician diagnosis in 95 9 of cases Keywords Ejection fraction Heart sound signal first and second sound Wavelet transform Electrocardiogram signal

رسول امير فتاحي، بهزاد نظري

مجتبي منصوري

سعيد صدري، محمدرضا احمدزاده