Decoding motor imagery loaded on steady-state somatosensory evoked potential based on complex task-related component...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Motor Imagery (MI) recognition is crucial for brain-computer interfaces (BCIs).
• The MI with steady-state somatosensory evoked potential (MI-SSSEP) paradigm offers higher accuracy than traditional MI.
• Existing algorithms often overlook MI-SSSEP signal characteristics, limiting performance and increasing false triggers.

Purpose of the Study:

• To develop an advanced algorithm for MI-SSSEP signal processing.
• To enhance the accuracy of motor imagery recognition.
• To reduce the false triggering rate in BCIs utilizing the MI-SSSEP paradigm.

Main Methods:

• Proposed the complex signal task-related component analysis (cTRCA) for spatial filtering of SSSEP signals.
• Validated cTRCA's effectiveness against traditional task-related component analysis (TRCA) using simulated and experimental MI-SSSEP data.
• Tested the algorithm's performance in identifying right-handed tasks against three interference tasks to assess false triggering rates.

Main Results:

• The cTRCA algorithm, combined with mutual information-based best individual feature (MIBIF) and minimum distance to mean (MDM), achieved an Area Under the Curve (AUC) of 0.89.
• This significantly outperformed the traditional Common Spatial Pattern (CSP) with Support Vector Machine (SVM) algorithm, which had an average AUC of 0.77 (p < 0.05).
• The cTRCA-based model reduced the false triggering rate from 38.69% to 20.74% compared to CSP+SVM (p < 0.001).

Conclusions:

• The study demonstrates that TRCA is sensitive to MI-SSSEP signal characteristics.
• Motor imagery tasks within the MI-SSSEP paradigm induce phase changes in evoked potentials.
• The cTRCA algorithm effectively leverages these phase changes, proving more suitable for the MI-SSSEP paradigm, improving motor imagery decoding, and reducing false triggers.