EMG map image processing for recognition of fingers movement.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Technology

Background:

• Electromyography (EMG) is a standard noninvasive technique for assessing muscle activity.
• Current methods for analyzing EMG data, especially for fine motor control, can be limited in precision.
• There is a need for advanced methods to interpret complex muscle activation patterns for assistive technologies.

Purpose of the Study:

• To develop and validate a new image processing technique for recognizing individual finger movements using EMG maps.
• To quantify the accuracy of this automated recognition system compared to expert clinical assessment.
• To explore the potential application of this technology in controlling assistive devices.

Main Methods:

• EMG signals were recorded using a 24-contact array electrode connected to a wireless digital amplifier.
• EMG maps were generated from the recorded signals to visualize muscle activity.
• An image processing algorithm was developed to detect and quantify high activity regions within the EMG maps.
• The system was tested on individuals without known motor impairments during specific finger movements.

Main Results:

• The developed method successfully identified temporal and spatial patterns in EMG maps corresponding to distinct finger movements.
• The automated recognition system achieved an average accuracy of 97.87% ± 0.92% when compared to expert clinician recognition.
• The results demonstrate the system's capability to differentiate subtle muscle activation patterns.

Conclusions:

• The novel image processing method based on EMG maps provides a highly accurate and reliable way to recognize individual finger movements.
• This technology has significant potential for real-time control of wearable assistive systems, such as hand prostheses and exoskeletons.
• The system's wearable nature and potential for microcomputer implementation make it suitable for practical assistive applications.