Shoulder Flexion Pre-Movement Recognition Through Subject-Specific Brain Regions to Command an Upper Limb Exoskeleton

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCIs) offer potential for motor rehabilitation.
• Accurate recognition of pre-movement states is crucial for effective BCI control.
• Electroencephalography (EEG) is a non-invasive method for capturing brain activity.

Purpose of the Study:

• To develop and evaluate two novel BCIs for shoulder pre-movement recognition.
• To compare manual versus automated (NMF) EEG channel selection strategies.
• To identify optimal frequency bands for discriminating pre-movement and rest states.

Main Methods:

• Two BCI approaches were implemented: manual EEG channel selection and subject-specific selection using Non-negative Matrix Factorization (NMF).
• Spatial features were extracted from filtered EEG signals using Riemannian covariance matrices.
• Linear Discriminant Analysis (LDA) was employed for classification of pre-movement and rest states.
• Experiments were conducted on 21 healthy subjects across various frequency bands.

Main Results:

• The automated BCI successfully identified EEG channels over the contralateral limb.
• Enhanced pre-movement recognition accuracy was achieved (ACC = 71.39 ± 12.68%, κ = 0.43 ± 0.25%).
• The study identified optimal frequency ranges for shoulder pre-movement detection.

Conclusions:

• The proposed BCIs demonstrate efficacy in recognizing shoulder pre-movement states.
• Automated EEG channel selection based on cortical relevance improves BCI performance.
• These findings suggest potential benefits for neuro-rehabilitation applications.