A multi-Kalman filter-based approach for decoding arm kinematics from EMG recordings
Hend ElMohandes1,2, Seif Eldawlatly3,4, Josep Marcel Cardona Audí5
1Center of Informatics Science, Nile University, Giza, Egypt.
Biomedical Engineering Online
|September 3, 2022
Summary
This study introduces a novel multi-Kalman filter approach for decoding arm kinematics from Electromyography (EMG) signals, enabling more natural prosthetic arm control. The method shows promise for reliable, continuous, and simultaneous movement decoding, potentially leading to subject-independent prosthetic systems.
Area of Science:
- Biomedical Engineering
- Rehabilitation Robotics
- Signal Processing
Background:
- Advancements in myoelectric prosthetics face challenges in reliability, naturalistic control, and computational demands.
- Existing systems often lack simultaneous, continuous control and are limited in movement scope.
Purpose of the Study:
- To propose and evaluate an Electromyography (EMG)-based multi-Kalman filter approach for decoding continuous and simultaneous arm kinematics.
- To decode elbow angle (θ) and wrist joint positions (X, Y) for enhanced prosthetic arm functionality.
Main Methods:
- Recorded arm kinematics and EMG signals from biceps, triceps, and deltoid muscles of ten subjects.
- Developed and applied a multi-Kalman filter algorithm to decode kinematic data from EMG signals.
- Assessed decoder performance using correlation coefficient (CC) and normalized root-mean-square error (NRMSE).
Main Results:
- Within-subject decoding achieved average CC of 0.68 (θ), 0.67 (X), and 0.64 (Y), with NRMSE of 0.21, 0.18, and 0.24, respectively.
- Cross-subject decoding yielded average CC of 0.61 (θ), 0.61 (X), and 0.48 (Y), with NRMSE of 0.23, 0.20, and 0.38, respectively.
Conclusions:
- The proposed EMG-based multi-Kalman filter approach demonstrates efficacy in decoding arm kinematics.
- Results suggest the potential for developing a subject-independent decoder for more accessible myoelectric prostheses.
