Two-stage binary classifier for neuromuscular disorders using surface electromyography feature extraction and...

Area of Science:

• Neurology
• Biomedical Engineering
• Signal Processing

Background:

• Needle electromyography is the standard for diagnosing neuromuscular disorders but can be invasive.
• Surface electromyography (sEMG) offers a non-invasive alternative for assessing neuromuscular function.
• Developing accurate sEMG-based diagnostic methods is crucial for improving patient care.

Purpose of the Study:

• To evaluate the efficacy of sEMG in classifying individuals with normal neuromuscular function, myopathy, and neuropathy.
• To develop and validate a machine learning model for sEMG-based diagnosis of neuromuscular disorders.
• To compare the diagnostic accuracy of a feature-selected two-stage binary classifier with other classification approaches.

Main Methods:

• sEMG signals were recorded during maximum voluntary isometric contractions and repetitive exercises.
• Feature extraction involved activity and frequency analysis of sEMG signals to assess muscle activity and fatigue.
• A two-stage binary classifier was implemented, with feature selection performed for each classification stage to optimize performance.

Main Results:

• The feature-selected two-stage binary classifier achieved a diagnostic accuracy of 86.9%.
• This accuracy surpassed that of a two-stage binary classifier without feature selection (82.3%) and a multi-class classifier (73.9%).
• Statistical feature selection significantly improved the classification performance for differentiating between normal, myopathy, and neuropathy groups.

Conclusions:

• sEMG analysis, combined with appropriate feature selection and classification algorithms, can accurately diagnose neuromuscular disorders.
• This approach demonstrates the potential of sEMG as a reliable, non-invasive tool for clinical diagnosis of myopathy and neuropathy.
• Further research can refine sEMG-based diagnostic systems for broader clinical application.