Decoding movement frequencies and limbs based on steady-state movement-related rhythms from noninvasive EEG

Area of Science:

• Neuroscience
• Neural Engineering
• Biomedical Engineering

Background:

• Decoding movements noninvasively using electroencephalography (EEG) is crucial for brain-computer interfaces (BCIs).
• While sensorimotor rhythm (SMR) is effective for limb decoding, it struggles with movement frequency decoding.
• Steady-state movement-related rhythm (SSMRR) is hypothesized to encode movement frequency.

Purpose of the Study:

• Investigate the spatial-spectral representation of SSMRR in EEG during voluntary movements.
• Assess the effectiveness of SSMRR for decoding movement frequencies and limbs.
• Explore SSMRR as a complement to SMR for enhanced BCI capabilities.

Main Methods:

• Examined frequency characteristics and spatial patterns of SSMRR during rhythmic finger movements.
• Utilized coherence analysis between EEG (sensor/source domain) and finger movements recorded by data gloves.
• Employed a fusion model combining spectral SNR and filter-bank common spatial pattern features for decoding.

Main Results:

• Coherence maps confirmed SSMRR, including the fundamental frequency (f0) and its first harmonic (f1), is encoded in the contralateral motor cortex.
• Achieved four-class decoding accuracies of 73.14% (externally paced) and 66.30% (internally paced) for movement frequency and limb across ten subjects.
• Top-performing subjects reached decoding accuracies of 87.21% and 80.44%.

Conclusions:

• Verified the EEG representation of SSMRR and its utility in decoding movement frequency and limb.
• Demonstrated that SSMRR, using spatial-spectral features, can be effectively decoded.
• Proposed SSMRR as a valuable addition to SMR for expanding decodable movement types in BCIs.