Residual one-dimensional convolutional neural network for neuromuscular disorder classification from needle electromyography signals with explainability

Jaesung Yoo, Ilhan Yoo, Ina Youn, Sung-Min Kim, Ri Yu, Kwangsoo Kim, Keewon Kim, Seung-Bo Lee

Abstract: Highlights•One-dimensional CNN was designed to classify neuromuscular disorders based on needle electromyography (nEMG) signals•nEMGNet, a one-dimensional CNN, captures raw nEMG signal characteristics at high accuracy and short inference time•Heterogeneous data structure for each subject is mitigated by divide-and-vote algorithm•Learned features of nEMGNet resemble the typical signal characteristics of neuromuscular disorders