Digital Signal Processing and Analysis of Cardiopulmonary Audio Using a Multi-Channel Stethograph System

A novel MATLAB based auscultation diagnosis program was developed to analyze audio from the heart, lungs and trachea (HLT) using digital signal processing techniques. The HLT audio were non-invasively measured using a custom built multi-channel stethograph system, which can record audio through a set of 16 microphone sensors. FIR filter, Hamming window and cross-correlation algorithm were used for signal preprocessing, data segmentation and feature extraction, respectively. In addition, a user interface that displays the processed audio in both time domain and frequency domain along with its energy spectrogram was also developed. The auscultation diagnosis program was able to filter and distinguish between good and bad quality audio, with an accuracy of 96%. The performance of the system is analyzed and presented in this paper.