Heart Sound Segmentation—An Event Detection Approach Using Deep Recurrent Neural Networks

摘要

Objective:nIn this paper, we accurately detect the state-sequencenfirst heart sound (S1)–systole–second heart sound (S2)–diastolen, i.e., the positions of S1 and S2, in heart sound recordings. We propose an event detection approach without explicitly incorporatingna priorininformation of the state duration. This renders it also applicable to recordings with cardiac arrhythmia and extendable to the detection of extra heart sounds (third and fourth heart sound), heart murmurs, as well as other acoustic events.nMethods:nWe use data from the 2016 PhysioNet/CinC Challenge, containing heart sound recordings and annotations of the heart sound states. From the recordings, we extract spectral and envelope features and investigate the performance of different deep recurrent neural network (DRNN) architectures to detect the state sequence. We use virtual adversarial training, dropout, and data augmentation for regularization.nResults:nWe compare our results with the state-of-the-art method and achieve an average score for the four events of the state sequence ofn${bf F}_{1}approx 96$n% on an independent test set.nConclusion:nOur approach shows state-of-the-art performance carefully evaluated on the 2016 PhysioNet/CinC Challenge dataset.nSignificance:nIn this work, we introduce a new methodology for the segmentation of heart sounds, suggesting an event detection approach with DRNNs using spectral or envelope features.