Toward Simultaneous Neurostimulation and Prosthetic Control: Real-Time Filtering of Amplitude-Modulated Stimulation...

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Engineering

Background:

• Restoring tactile feedback is crucial for intuitive prosthetic control.
• Electrical artifacts from nerve stimulation distort electromyographic (EMG) signals, impairing prosthetic decoding accuracy, especially in implanted systems.

Purpose of the Study:

• To develop and evaluate a real-time filtering algorithm for subtracting stimulation artifacts from implanted EMG signals.
• To improve prosthetic control and decoding accuracy during simultaneous sensory feedback stimulation.

Main Methods:

• A real-time adaptive filtering algorithm using sample-wise polynomial regression based on stimulation amplitude was developed.
• The algorithm was implemented on an embedded controller and tested in a participant with a long-term neuromusculoskeletal interface.
• Offline and real-time motion classification tasks were performed under fixed and amplitude-modulated stimulation.

Main Results:

• The filter effectively restored EMG feature distributions in offline tests.
• Real-time movement completion rates improved from 20% to 60% (fixed stimulation) and 32% to 44% (modulated stimulation).
• Significant recovery of decoding performance was achieved during dynamic stimulation.

Conclusions:

• The developed filtering algorithm effectively mitigates stimulation artifacts in implanted EMG signals.
• This technology enables robust, closed-loop control in bidirectional prostheses, addressing a key barrier in prosthetic development.
• The findings provide a foundation for advanced prosthetic systems integrating sensory feedback and motor control.