Classification of EMG signals using PCA and FFT

Area of Science:

• Biomedical Engineering
• Neuroscience
• Machine Learning in Medicine

Background:

• Electromyography (EMG) signals are crucial for diagnosing neuromuscular disorders.
• Analyzing complex EMG data efficiently is essential for clinical applications.
• Fast Fourier Transform (FFT) and Principal Component Analysis (PCA) offer potential for EMG signal processing.

Purpose of the Study:

• To apply FFT analysis and PCA to EMG signals for improved interpretation.
• To compare the diagnostic performance of Multilayer Perceptron (MLP) and Support Vector Machine (SVM) classifiers.
• To evaluate the efficacy of machine learning models in distinguishing normal, neuropathy, and myopathy cases.

Main Methods:

• EMG signals from 59 patients (19 normal, 20 neuropathy, 20 myopathy) were analyzed using FFT.
• PCA was employed to reduce the dimensionality of FFT coefficients for efficient data handling.
• Reduced PCA coefficients were used to train and test MLP and SVM classification models.

Main Results:

• Support Vector Machine (SVM) demonstrated a high level of accuracy in diagnosing neuromuscular disorders.
• SVM exhibited superior test performance compared to the Multilayer Perceptron (MLP) model.
• The combined FFT and PCA approach facilitated data calculation and storage.

Conclusions:

• SVM is a highly effective tool for the automated diagnosis of neuromuscular disorders using processed EMG data.
• PCA-based feature reduction enhances the efficiency of EMG signal analysis for machine learning.
• This study validates the potential of machine learning, particularly SVM, in clinical neurophysiology.