Related Experiment Video
Updated: Apr 28, 2026
Capturing Dynamic Finger Gesturing with High-resolution Surface Electromyography and Computer Vision
Published on: March 28, 2025
Fuzzy MUAP recognition in HSR-EMG detection basing on morphological features
Kamel Mebarkia1, Raїs El'hadi Bekka2, Aicha Reffad3
1Université de Sétif 1, Faculty of Technology, Department of Electronics, LIS Laboratory, 19000 Sétif, Algeria; Department of Rehabilitation and Prevention Engineering, Institute of Applied Medical Engineering, RWTH-Aachen University, 52074 Aachen, Germany.
This study developed a fuzzy classifier to distinguish isolated from overlapped motor unit action potentials (MUAPs). The system achieved high accuracy, demonstrating its effectiveness in analyzing electromyography signals.
Area of Science:
- Biomedical Engineering
- Signal Processing
- Neuromuscular Physiology
Background:
- Motor Unit Action Potentials (MUAPs) exhibit unique patterns influenced by recording techniques.
- Differentiating isolated MUAPs from overlapped ones is crucial for accurate electromyography (EMG) analysis.
- Novel morphological features are needed for robust MUAP detection.
Purpose of the Study:
- To develop and evaluate a fuzzy classifier for distinguishing Laplacian-detected isolated MUAPs from overlapped MUAPs.
- To utilize novel morphological features for improved MUAP recognition.
- To assess the classifier's performance across different muscles and noise levels.
Main Methods:
- Development of an 'if-then' fuzzy rule-based classifier.
- Utilizing morphological features of Laplacian-detected MUAPs.
- Training and testing the classifier on EMG data from abductor pollicis brevis (APB), first dorsal interosseous (FDI), and biceps brachii (BB) muscles.
- Optimization of fuzzy rules using genetic algorithms (GA).
- Evaluation using synthetic signals and varying signal-to-noise ratios (SNR).
Main Results:
- The fuzzy classifier achieved 95.03% accuracy in recognizing isolated MUAPs.
- Performance improved to 97.8% after rule optimization with genetic algorithms.
- The classifier maintained acceptable performance down to an SNR of 20 dB for isolated MUAPs (detection probability 0.96).
- Recognition of overlapped MUAPs was less sensitive to noise (detection probability ~0.8), with misrecognition primarily due to synchronization and small overlap degrees.
Conclusions:
- The proposed fuzzy classifier effectively distinguishes isolated MUAPs from overlapped ones using novel morphological features.
- Genetic algorithm optimization significantly enhances classifier performance.
- The method is robust to noise, particularly for isolated MUAPs, making it suitable for clinical EMG analysis.
