Two-Source Validation of Online Surface EMG Decomposition Using Progressive FastICA Peel-Off

Area of Science:

• Biomedical Engineering
• Neuroscience
• Signal Processing

Background:

• Online decomposition of surface electromyogram (SEMG) lacks validation on real experimental data due to unknown motor unit (MU) activities.
• Previous studies relied on simulated signals, limiting comprehensive assessment of SEMG decomposition methods.

Purpose of the Study:

• To conduct a comprehensive validation of online SEMG decomposition using simultaneously recorded intramuscular EMG (IEMG) and high-density SEMG signals.
• To assess the accuracy and reliability of online SEMG decomposition by comparing it against a ground-truth reference derived from IEMG.

Main Methods:

• A two-source validation approach using simultaneous IEMG and high-density SEMG recordings.
• Decomposition of IEMG using a simplified Progressive FastICA Peel-off (PFP) method to establish ground-truth MU spike trains.
• Offline extraction of MU separation vectors from initial SEMG signals for online MU spike train extraction.

Main Results:

• A total of 549 MUs from SEMG and 92 MUs from IEMG were identified in 5 healthy subjects.
• All MUs decomposed from IEMG were successfully matched with MUs from online SEMG decomposition.
• The average matching rate for common firing events in the online stage was high at (96 ± 1)%.

Conclusions:

• The study provides robust validation for online SEMG decomposition using experimental data.
• Separation vectors effectively track the same MU continuously and precisely in experimental SEMG signals.
• This research offers a more comprehensive validation perspective for online SEMG decomposition.