Identification of isometric contractions based on High Density EMG maps

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Technology

Background:

• Accurate identification of motion intention and muscle activation is crucial for human-machine interfaces like prostheses and rehabilitation devices.
• Existing pattern recognition methods using surface electromyography (sEMG) often struggle with variations in muscle strength and electrode placement.
• High Density surface Electromyography (HD-EMG) offers richer spatial information for improved motion intent and muscle load distribution analysis.

Purpose of the Study:

• To automatically identify four isometric motor tasks of the forearm (flexion-extension, supination-pronation).
• To differentiate between low-to-medium voluntary contraction levels.
• To develop a classifier robust to variations in muscle strength and spatial distribution using HD-EMG.

Main Methods:

• Acquisition of monopolar HD-EMG maps from five upper-limb muscles in healthy subjects during isometric contractions.
• Development of a novel classifier combining two-step linear discriminant analysis with spatial distribution and intensity-based features from HD-EMG maps.
• Utilizing an "isolated masses" method for calculating spatial distribution features.

Main Results:

• The proposed classifier successfully identified isometric motor tasks and differentiated contraction levels, even at 10% of Maximum Voluntary Contraction (MVC).
• Combining intensity-based features with spatial distribution features (using the "isolated masses" method) significantly improved classifier performance.
• The method demonstrated potential for estimating motion intention and effort in the low-effort range relevant for patients with neuromuscular disorders.

Conclusions:

• HD-EMG, when analyzed with advanced feature extraction and classification techniques, provides a robust method for identifying motion intention and effort.
• The developed classifier shows promise for applications in advanced prosthetic control, assistive devices, and personalized rehabilitation programs.
• This approach can enhance the functionality and user experience of human-machine interfaces in various biomedical applications.