Combining electromyographic and electrical impedance data sets through machine learning: A study in D2-mdx and...

Area of Science:

• Biomedical Engineering
• Neuroscience
• Muscle Physiology

Background:

• Needle impedance-electromyography (iEMG) offers a novel method to simultaneously measure muscle active (EMG) and passive (EIM) electrical properties.
• Muscular dystrophy models present a challenge for accurate diagnosis and monitoring.

Purpose of the Study:

• To assess the efficacy of combining multifrequency EMG and EIM data using machine learning (ML) for discriminating between D2-mdx muscular dystrophy mice and wild-type (WT) controls.
• To explore a novel approach for characterizing the full electrical properties of skeletal muscle.

Main Methods:

• iEMG data were acquired from quadriceps muscles of D2-mdx and WT mice under varying anesthesia levels.
• EMG data underwent Fourier transformation to obtain power spectra, while EIM data were collected concurrently.
• A nested machine learning approach, specifically random forest, was employed to classify healthy versus disease states using EIM, EMG, and combined datasets.

Main Results:

• Electrical impedance myography (EIM) data alone achieved 93.1% accuracy in differentiating D2-mdx from WT mice.
• Electromyography (EMG) data alone achieved 75.6% accuracy.
• Combining EIM and EMG data resulted in 92.2% accuracy, showing no significant improvement over EIM alone.

Conclusions:

• The study successfully demonstrated an ML-based approach for integrating EIM and EMG data from a novel iEMG needle.
• While the combination of EIM and EMG did not outperform EIM alone in this specific dataset, the methodology represents a novel way to analyze comprehensive muscle electrical characteristics.
• This approach holds potential for future investigations into muscle diseases.