Heart Murmur Detection of PCG Using ResNet with Selective Kernel Convolution

Aims: Heart murmur detection plays a crucial role in the early diagnosis of congenital and acquired heart diseases in children. This study aimed to construct a deep neural network architecture for detecting heart murmurs from PCG recordings. Methods: To learn effective features, we constructed a ResNet with selective kernel convolution (SK-Conv). The SK-Conv was embedded into each ResBlock, which adaptively captures multi-scale features using convolution filters of different kernel sizes and applies a channel attention module (similar to Squeeze-and- Excitation) to emphasize the representation of important features. Our model was trained on record-level data and validated on patient-level data. The final classification result will be obtained by our proposed multi-level voting rules. Results: In the official phase, our model(fty+_h) was evaluated in a hidden test set. Specifically, the proposed classifier ranked 33rd in the official murmur detection task with a weighted accuracy and a cost of 0.525 and 22654, respectively. In addition, our method ranked 35th in the official clinical outcome identification task with a weighted accuracy and a cost of 0.5 and 15,982, respectively.