Deep learning-based algorithm versus physician judgement for myopathy and neuropathy diagnosis based on needle electromyography findings

Abstract

Electromyography is a valuable diagnostic tool for diagnosing patients with neuromuscular diseases; however, it has possible drawbacks including diagnostic accuracy and a time- and effort-intensive process. To overcome these limitations, we developed a deep learning-based electromyography diagnosis system and compared its performance with that of six physicians. This study included 58 participants who underwent electromyography and were subsequently confirmed to have myopathy or neuropathy or to be in a normal state at single tertiary centre. We developed a one-dimensional convolutional neural network and Divide-and-Vote algorithms for diagnosing patients. Diagnostic results from our deep learning model were compared with those of six physicians with experience in performing and interpreting electromyography. The accuracy, sensitivity, specificity, and positive predictive value of the deep learning model were 0.875, 0.820, 0.904, and 0.820, respectively, whereas those of the physicians were 0.694, 0.537, 0.773, and 0.524, respectively. The area under the receiver operating characteristic curves of the deep learning model was also better than those of the averaged results of the six physicians. Thus, deep learning could play a key role in diagnosing patients with neuromuscular diseases.