Neural Noise

Lower-limb motor imagery EEG BCI in knee pain patients

2 min read

  • Motor imagery classification was enhanced on a large lower-limb EEG dataset in knee pain patients.1
  • Methods transfer to lower-limb and pain cohorts; dataset size supports robustness for BCI.1 1

Weekly enrichment (2026-07-20)

  • The underlying paper, “Enhancing classification of a large lower-limb motor imagery EEG dataset for BCI in knee pain patients,” was published in Scientific Data (Nature portfolio), DOI 10.1038/s41597-025-05767-2 (corresponding author Chongwen Zuo).2 3
  • The authors describe it as the first large-scale, standardized lower-limb motor imagery (MI) EEG dataset for knee pain patients: 30 knee-joint-pain patients, 150 sessions, and 15,000 trials (30 patients × 5 sessions × 100 trials).2 4
  • Each participant performed MI experiments every two days over five independent sessions, imagining left-leg and right-leg flexion/extension based on on-screen cues under experimenter supervision.2
  • MI tasks elicited significant event-related desynchronization/synchronization (ERD/ERS), analyzed particularly over the central sensorimotor channels C3 and C4, with sensorimotor rhythms concentrated in the alpha (8–15 Hz) and beta (15–30 Hz) bands.2 3
  • The proposed Optimal Time-Frequency Window Riemannian Geometric Distance (OTFWRGD) algorithm reached 86.41% average classification accuracy, substantially outperforming CSP+LDA (51.43%), FBCSP+SVM (55.71%), and the deep-learning EEGNet (76.21%).2 4
  • The dataset adheres to EEG-BIDS standards and is released openly on Figshare (DOI 10.6084/m9.figshare.28740260), including raw and preprocessed EEG, stimuli, and analysis code.3 4
  • A key clinical implication is that, although chronic knee pain can alter cortical plasticity, the data demonstrate preserved MI capability in these patients—supporting the feasibility of MI-BCI for lower-limb neurorehabilitation.4
  • The finding is consistent with prior knee-MI work showing foot-area mu-rhythm desynchronization (most reactive component ~8.8 ± 0.5 Hz) with contralateral dominance at central electrodes, reinforcing sensorimotor rhythms as usable BCI control signals for leg movement.5

Footnotes

  1. https://news.google.com/rss/articles/CBMiX0FVX3lxTE96ZTJnSy1zNEcxMVhjZkFGV0hSTUp6YnpQMzBUeVprbXd1em1CenBLYXNCSEk0ZzBqUXVXbnlGckUwQkNGNmRQUXhIWXQ4OFUwN1kzUC16eFVLcFd4RUs4?oc=5 ↩ ↩2 ↩3

  2. https://www.nature.com/articles/s41597-025-05767-2 ↩ ↩2 ↩3 ↩4 ↩5

  3. https://pmc.ncbi.nlm.nih.gov/articles/PMC12368069/ ↩ ↩2 ↩3

  4. https://doi.org/10.6084/m9.figshare.28740260 ↩ ↩2 ↩3 ↩4

  5. https://doi.org/10.1109/ascc.2017.8287519 ↩

Graph View