IEEE transactions on computational biology and bioinformatics · 2026
Deep Learning-Based Volition Detection and Action Potential Extraction for Fully Automated Diagnosis of Neuromuscular Disease Using Needle Electromyography Signals.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Chung Soomin, Yoo Ilhan, Lee Jinkyu, Kim Dongmin, Kim Kwangsoo, Kim Keewon, Lee Seung-Bo
Аннотация
OBJECTIVE: This study aimed to develop a deep learning-based volition-detection model to automate the diagnostic process and improve neuromuscular disease classification using needle electromyography (nEMG) signals.
METHODS: The model was developed using 376 nEMG signals from 57 subjects and independently evaluated on an external dataset of 751 nEMG signals from 115 subjects at a single tertiary medical center. The proposed model directly processed raw nEMG signals to automatically extract volition signals (motor unit action potentials), eliminating the need for physician-dependent preprocessing.
RESULTS: The optimal segment length for volition detection was 0.060 s, and nEMGNet demonstrated the best overall performance. Model-extracted volition signals yielded higher segment-wise classification performance than raw nEMG signals, with area under the receiver operating characteristic curve (AUROC) values of 0.806 for myopathy, 0.819 for neuropathy, and 0.819 for normal cases, compared with 0.761, 0.728, and 0.733, respectively. At the patient-wise level, AUROC improved by 20.10% for myopathy, 23.15% for neuropathy, and 23.15% for normal cases compared with physician-detected volition data.
CONCLUSION: The proposed volition-detection model significantly enhances neuromuscular disease classification performance while enabling a fully automated and less labor-intensive diagnostic workflow.
METHODS: The model was developed using 376 nEMG signals from 57 subjects and independently evaluated on an external dataset of 751 nEMG signals from 115 subjects at a single tertiary medical center. The proposed model directly processed raw nEMG signals to automatically extract volition signals (motor unit action potentials), eliminating the need for physician-dependent preprocessing.
RESULTS: The optimal segment length for volition detection was 0.060 s, and nEMGNet demonstrated the best overall performance. Model-extracted volition signals yielded higher segment-wise classification performance than raw nEMG signals, with area under the receiver operating characteristic curve (AUROC) values of 0.806 for myopathy, 0.819 for neuropathy, and 0.819 for normal cases, compared with 0.761, 0.728, and 0.733, respectively. At the patient-wise level, AUROC improved by 20.10% for myopathy, 23.15% for neuropathy, and 23.15% for normal cases compared with physician-detected volition data.
CONCLUSION: The proposed volition-detection model significantly enhances neuromuscular disease classification performance while enabling a fully automated and less labor-intensive diagnostic workflow.