A multi-paradigm EEG dataset for studying upper limb rehabilitation exercises

Area of Science:

• Neuroscience
• Rehabilitation Engineering
• Biomedical Signal Processing

Background:

• Stroke survivors often face lasting upper limb motor deficits.
• Brain-computer interface (BCI) technologies show promise for rehabilitation.
• Existing EEG datasets lack multi-paradigm comparisons from the same subjects, hindering progress.

Purpose of the Study:

• To construct a comprehensive EEG dataset encompassing diverse upper limb rehabilitation paradigms.
• To facilitate comparative analysis of neural mechanisms across different rehabilitation strategies.
• To support the development of optimized BCI-based rehabilitation protocols.

Main Methods:

• Recruited 28 healthy subjects for EEG data collection.
• Recorded electroencephalography (EEG) signals during six distinct upper limb rehabilitation paradigms.
• Included actions like grasping and releasing with single or both hands.
• Provided both raw and preprocessed EEG data (bandpass filtering, artifact removal).

Main Results:

• Successfully collected a novel multi-paradigm EEG dataset from healthy subjects.
• The dataset includes detailed information on various upper limb motor actions.
• Offers both raw and processed EEG signals for diverse analytical approaches.

Conclusions:

• The created EEG dataset is a valuable resource for BCI research.
• Enables direct comparison of neural activity across different rehabilitation paradigms.
• Will accelerate the development of more effective stroke rehabilitation strategies.