Cardinality with spatial filter for myoelectric pattern recognition feature extraction
Abstract
This research explores improving the control of powered prostheses for amputees using myoelectric pattern recognition (PR). A key challenge lies in extracting reliable features from electromyogram (EMG) signals, which are often processed independently per channel, neglecting inter- muscle interactions. To address this, the study introduces range spatial filtering (RSF), a novel spatial filtering method. RSF creates logical combinations of EMG channels, effectively increasing the data available for feature extraction and accommodating varying muscle contributions during movement. This is combined with traditional cardinality time- domain (TD) feature extraction, leveraging its cost-effectiveness to enhance classification performance. The proposed technique was tested on EMG data from nine trans-radial amputees, capturing six grasp and finger motions at three force levels. Results demonstrate a significant reduction in classification error rates (almost 10%, p < 0.001) compared with alternative approaches. The study highlights the potential of RSF to improve the reliability and accuracy of myoelectric interfaces, paving the way for more effective real-world implementation of EMG-based prostheses.
Keywords
myoelectric control, electromyography, prosthetics, pattern recognition, spatial filtering, classification accuracy
Document Type
Book Chapter
Date of Publication
1-1-2026
Publication Title
Innovations in Electrical and Mechanical Engineering for A Sustainable Future
Publisher
Taylor & Francis
School
School of Engineering
Copyright
subscription content
First Page
105
Last Page
116
Comments
Al Taee, A. A., Khushaba, R., Zia, T., & Al-Jumaily, A. (2026). Cardinality with spatial filter for myoelectric pattern recognition feature extraction. In S. M. Sulthan, K. H. Law, & A. R. Alao (Eds.), Innovations in electrical and mechanical engineering for a sustainable future (pp. 1–12). Taylor & Francis. https://doi.org/10.1201/9781779644084-9