Automatic Phonocardiography Analysis Using Discrete Wavelet Transform

This paper presents an intelligent algorithm for analyzing the heart sound signal using multi-resolution analysis based on discrete wavelet transform (DWT). This signal processing technique can not only compress and encrypt the phonocardiogram (PCG) signal but also reduces the noises and the murmurs from the PCG signal. This information can assist the cardiologists for taking appropriate actions to diagnosis for the initial stage of the cardiovascular dis- order. We have presented an efficient data compression algorithm based on DWT, energy packing efficiency (EPE), and run-length encoding (RLE) that can compress the signal about 93.70% without losing any pathological information. Further, this paper explores an end-to-end encryption technique using the compression algo-rithm to maintain patient confidentiality. After the compression of the PCG signal, the noises and the murmurs are effectively separated from the signal by reusing DWT. The performance of this algorithm has been evaluated by using 5 normal and 18 abnormal PCG signals available in the University of Michigan Heart Sound and Murmur Library. This method can be used in real time for the remote monitoring of the heart sound.