Heart Murmur Detection

This project demonstrates a novel selective filtering approach to enhance the detection of heart murmurs in phonocardiography. By focusing on isolating the frequencies associated with murmurs while minimizing interference from the dominant heartbeat sounds, the method allows for clearer analysis of cardiac conditions. The systematic workflow includes data acquisition, spectrogram construction, noise reduction, and peak identification, enabling the differentiation of various types of murmurs—systolic, diastolic, and continuous. The results indicate a significant improvement in the audibility of heart murmurs, thus supporting more accurate diagnoses.

I approached the project in the following way. First I would extract the data from the stethoscope over bluetooth. Then I would run a python program to record this data and save it as a .wav file. After this the data got sent to processing, and it's spectrogram was plotted. A peak finding algorithm would then find the peaks of intensity(heart beats) and their time locations would be recorded. After that a filter is applied at those time locations, and the rest of the audio is amplified. Then all graphs and sound recordings of the processed and preprocessed data is sent to a server for the doctor to view.

This project was a massive learning experience for me. I advanced my knowledge in digital signal processing, and really got my python fundamentals down right. This project was also done alone, so I gained a lot of insight into leading a project solo and the things that come with that. I also learned a great deal on how the heart works and the things that we should be listening for when it comes to phonocardiography.