Real-time adaptive cancellation of TENS feedback artifact on sEMG for prosthesis closed-loop control

Area of Science:

• Biomedical Engineering
• Neuroprosthetics
• Signal Processing

Background:

• Prosthetic hands aim to restore function using bio-signals like Surface Electromyogram (sEMG) and sensory feedback via Transcutaneous Electrical Nerve Stimulation (TENS).
• TENS used for sensory feedback can introduce "Artifacts" that interfere with sEMG signals, degrading prosthetic hand control.
• Effective artifact removal is crucial for reliable intention estimation in advanced prosthetic systems.

Purpose of the Study:

• To develop and validate an adaptive artifact removal method for TENS interference in sEMG signals.
• To improve the performance of prosthetic hand intention estimation by enhancing sEMG signal quality.
• To demonstrate the efficacy of the proposed method in both offline and online experimental settings.

Main Methods:

• An adaptive artifact removal method employing a modified least-mean-square adaptive filter was proposed.
• The filter utilizes previous artifact means as reference signals and incorporates prior TENS system information.
• Temporal separation was used for artifact discrimination to enhance removal efficiency, validated on four sEMG signals.

Main Results:

• The adaptive filtering method significantly increased Signal-to-Noise Ratio (SNR) by an average of 10.3dB.
• Real-time prosthetic hand control experiments (Target Reaching Experiment - TRE) showed performance recovery to levels without TENS interference.
• The method demonstrated high artifact removal efficiency across variable conditions and online simulations.

Conclusions:

• The proposed adaptive artifact removal method effectively cancels TENS-induced artifacts from sEMG signals.
• This technique substantially improves signal quality, crucial for accurate intention estimation in TENS-based prosthetic feedback systems.
• The findings pave the way for more reliable and functional neuroprosthetic devices.