521273S Biosignal Processing Lab – V Spectral Analysis of Heart Sounds Objective Your task is to analyze the spectral content of the heart sounds of five different patients. Two of the patients (pat1 and pat2) have normal heart sounds, whereas two (pat3 and pat4) have a ventricular septal defect (a hole between the ventricles which causes blood flow from the left to the right ventricle during systole) causing a systolic murmur (=abnormal heart sound). One patient (pat5) has aortic stenosis, in which the aortic valve does not open properly due to its stiffened leaflets. This results in constrained ejection of blood into the aorta during ventricular systole and, again, a systolic murmur. Ventricular systole means the time period, when ventricles are contracting and blood pressure is reaching its maximum value in aorta. Implementation Data: The data is given in one file containing the phonocardiograms (PCG) and electrocardiograms (ECG) of all five patients. The sampling rate of the signals is 1000 Hz. You can listen to the PCGs using the Matlab function ‘sound’ in case you are interested. Some of the data files have artifacts at the beginning and/or ending. Exclude these parts from the analysis. Segmentation of systolic segments from PCG: 1) Use given QRSdetection.m and detectPeaks function within it to detect QRS onsets from ECG signals. QRSdetection is based on the Pan-Tompkins algorithm presented in the course book and the exercise 3. Note, that the sampling frequency of given data and codes is not the same and thus, make necessary resampling in your code. 2) Segment the systolic parts of the PCG signals by selecting a window of duration 300 - 350 ms starting from the beginning of each QRS complex. (The window should include the beginning of the S1 which is normal systolic sound occurring after the beginning of QRS complex and possible systolic murmur, but not the S2 sound which is normal diastolic sound occurring during/after T wave.) Spectral Analysis: Compute the power spectral density (PSD) for each PCG segment using the Welch’s averaged periodogram method (function ‘pwelch’). Furthermore, calculate the average PSD and its mean frequency for each patient. Requirements As a result you should have a figure with 15 (5x3) subplots. Each row of this figure should include following plots from each patient: • the first systolic PCG segment • PSD of this segment • average PSD computed using all PCG segments In addition, you should have the mean frequencies for all patients. This lab exercise is based on the material provided in http://people.ucalgary.ca/~ranga/enel563/. The material is used with the permission of Professor R. M. Rangayyan.