Abstract | ||
---|---|---|
Heart sound auscultation is a powerful and noninvasive technique for cardiac examinations. In this study, we propose a novel method for phonocardiogram (PCG) signal processing to enable automatic systolic murmur diagnosis. We formalize a series of analytical stages in this system and investigate the effectiveness of each step with real patient data. Firstly, in the heart sound segmentation step, a novel envelope is generated from short-time Fourier transform (STFT) of PCG signal to determine the positions of the first and second heart sounds. Meanwhile, the noisy cardiac cycle can be detected and deleted in this step. Then, we extract a 6 dimension feature, which can indicate the possibility of the occurrence of pathologic murmurs, by clustering the systolic STFT frames. Finally, a support vector machine (SVM) based classifier is trained to distinguish between heart sounds with and without murmurs. The proposed diagnostic system is evaluated by repeated random sub-sampling in an open PCG database and the results show a relatively high accuracy, sensitivity and specificity of 93.91±6.51%, 93.00±7.48% and 100% respectively for the detection of PCGs with systolic murmur.
|
Year | DOI | Venue |
---|---|---|
2019 | 10.1145/3358331.3358368 | Proceedings of the 2019 International Conference on Artificial Intelligence and Advanced Manufacturing |
Keywords | Field | DocType |
aortic stenosis (AS), heart sound, machine learning, mitral regurgitation (MR), signal processing | Phonocardiogram,Heart murmur,Internal medicine,Cardiology,Medicine | Conference |
ISBN | Citations | PageRank |
978-1-4503-7202-2 | 0 | 0.34 |
References | Authors | |
0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Wenjing Wei | 1 | 0 | 0.34 |
Ge Zhan | 2 | 0 | 2.03 |
Xun Wang | 3 | 0 | 0.68 |
Pengyuan Zhang | 4 | 50 | 19.46 |
Yonghong Yan | 5 | 656 | 114.13 |