A hybrid method for real-time stimulation artefact removal during functional electrical stimulation with time-variant parameters

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Technology

Background:

• Functional electrical stimulation (FES) is crucial for restoring motor function but is often corrupted by stimulation artefacts (SAs).
• Real-time extraction of volitional surface electromyography (sEMG) during FES is challenging due to time-variant parameters and signal interference.
• Accurate sEMG signals are vital for closed-loop FES control and understanding neural pathways.

Purpose of the Study:

• To propose and validate a hybrid hardware-software method for real-time SA removal and sEMG extraction during dynamic FES.
• To develop a novel database-based Gram-Schmidt (DBGS) algorithm for efficient SA de-artefacting.
• To assess the system's performance in able-bodied volunteers and individuals with stroke.

Main Methods:

• A hardware front-end (DFE) with fast recovery and blanking (7.6 ms) was designed to prevent saturation.
• A database-based Gram-Schmidt (DBGS) algorithm utilizing high-similarity SA templates was employed for de-artefacting.
• The prototype system was tested on eight able-bodied volunteers and three individuals with stroke.

Main Results:

• The 6th-order DBGS achieved an average stimulus artefact attenuation of 12.77 ± 0.85 dB and an SA index of 1.82 ± 0.37 dB.
• Correlation coefficients between clean and extracted sEMG reached 0.84 ± 0.33 dB, outperforming existing methods.
• Significant sEMG-torque correlations were observed (0.78 ± 0.05 for able-bodied, 0.48 ± 0.11 for stroke patients).

Conclusions:

• The proposed hybrid method effectively removes SAs and extracts sEMG in real time during dynamic FES.
• The DBGS algorithm demonstrates superior performance compared to traditional methods for SA removal.
• This technology holds promise for advancing real-time control in FES applications.