EMG-Based Intraoperative Neuromonitoring Using Advanced Machine Learning Approaches

Abstract

Intraoperative neuromonitoring (IONM) plays a critical role in minimizing nerve damage during surgeries by providing real-time feedback on neural integrity. This study evaluated models associated with deep learning and machine learning models for electromyography classification of signal during intraoperative neuromonitoring (IONM). The CNN-LSTM model achieved the highest accuracy (85.2%), outperforming traditional models like KNN (53%), RF (62%), and CNN (76%). This demonstrates the degree to which the CNN-LSTM model can gain insight into temporal and spatial dependencies throughout the EMG signals, which makes it optimal for real-time classification in IONM applications. This implies that deep learning techniques can improve surgical procedures' safety and efficacy.