HS Stroke: Upper-Limb Motor Imagery EEG from 14 Chronic Stroke Patients

56,000 motor imagery EEG trials were recorded from 14 chronic stroke patients over 20 sessions each. Jian Luo released this dataset on figshare in 2026, organized according to the EEG-BIDS standard. It includes pre- and post-training clinical measures like FMA, sEMG, and MEPs to assess motor function and neural activity.

Use Cases

• Train binary classifiers for left- vs. right-hand motor imagery based on EEG signals.
• Investigate correlations between EEG features and clinical motor function scores (FMA).
• Benchmark state-of-the-art decoding algorithms for motor imagery BCI systems.
• Study longitudinal neurophysiological changes in stroke patients across 20 training sessions.

Strengths

• Contains 56,000 labeled motor imagery EEG trials, providing substantial data for model training.
• Includes complementary clinical and neurophysiological measures (FMA, sEMG, MEPs) for multi-modal analysis.
• Data is organized according to the EEG-BIDS standard, which suggests structured and documented formatting.
• Achieved an average binary classification accuracy of 82.65% in baseline analysis, indicating signal quality.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect bias inherent to the specific cohort of 14 chronic stroke patients from a single study.

Provenance

Source

figshare, author Jian Luo.

Collection Method

EEG recordings from 14 chronic stroke patients performing left- and right-hand motor imagery tasks over 20 sessions.

Freshness

Last updated 2026-04-24 18:36:47; freshness should be verified.