A phonocardiogram(PCG) measures the heart sounds produced by the turbulent flow of blood in and out of the heart and the movement of valves that regulating this flow. Automatic analysis of PCG provides useful clinical information about heart operations for early detection and diagnosis of heart disorders. In this paper we design an algorithm to detect and classify the heart sound (PCG) signal into normal and abnormal sounds by using fuzzy inference system (FIS). The algorithm consisted of four stages, they are: preprocessing stage, segmentation stage, feature extraction stage and finally classification stage. In the preprocessing stage, PCG signals were first down sampled, and filtered by adaptive wavelet-based sub-level tracking. Next, the filtered PCG signal was segmented using envelop based segmentation using Shannon energy. Thereafter, zero crossing rate and root mean square features were extracted from the Shannon envelop and were used as inputs to FIS. Proposed algorithm reached 100% accuracy, and it was showed that proposed algorithm is effective for detection of abnormal heart operations.
Volume 12 | Issue 3
Pages: 385-392
DOI: 10.5373/JARDCS/V12I3/20201205