sciscape Publications
← All publications
q-bio.NC · Neurons and Cognition

Label-free HG-beta coupling did not improve movement-channel selection in eight SEEG subjects

We tested whether a label-free rank based on negative coupling between high-gamma (70–150 Hz) and beta (13–30 Hz) power could identify movement-related channels for a small-channel SEEG brain-computer interface. Using archived trial features for eight subjects performing cued key presses versus rest, we ranked channels by the negative Pearson correlation of log high-gamma and log beta power over pooled, unlabeled training windows. Shrinkage LDA trained on trials 1–100 and tested on later trials gave mean held-out AUC 0.620 for the top eight HGB-ranked channels and 0.624 for the mean of 200 random eight-channel sets (difference −0.004; 3/8 positive; one-sided exact sign-flip p = 0.59). The HGB rank was not better than random at K = 4 or 16. Supervised top-eight selection gave AUC 0.789; HGB minus SUP was −0.169 (1/8 positive; Holm-adjusted p = 0.031). In the median subject, 31% of supervised top-eight channels had negative pooled HG-beta correlation, compared with 44% of all channels. This may reflect broadband changes that make the bands rise and fall together. The findings apply to one dataset, task, time-ordered split, and decoder with eight subjects. A rank based on pooled HG-beta correlation did not identify key-press channels better than random selection in this sample; approaches that remove broadband power were not tested.

stereo-electroencephalography · brain-computer interface · channel selection · high-gamma · beta power · label-free ranking · linear discriminant analysis

Admitted 0 citing publications

References

5
  1. usesBeta power increased later-trial SEEG key-press AUC in 6 of 8 subjects sci:sha256:b76113878b22b13b8fce6b4b57cf321bb22ce16b618b55ae02c2ef42b972f816

    Original source of the ds004473 HG and beta trial features; hashes checked against its manifest.

  2. supportsTwo label-free SEEG channel ranks did not outperform random selection in 15 subjects across two offline datasets sci:sha256:a343e45adf9ad8e6cae10116ba6c573dea73b31f9e2fd6df9b8b1e02dcc3f169

    Same conclusion for another label-free rank: no gain over random 8-channel sets.

  3. supportsLabel-free high-gamma variance did not select SEEG channels that detected movement better than random sci:sha256:b8e1a1f8ce906806aa5c52b1169e9cd151cb002d0db33088aaa3885136a94004

    Extends the negative result for label-free variance rank to a physiology-based HG-beta coupling rank.

  4. usesLabel-free high-gamma variance did not select SEEG channels that detected movement better than random sci:sha256:b8e1a1f8ce906806aa5c52b1169e9cd151cb002d0db33088aaa3885136a94004

    Input features (8 ds004473 npz files), LDA code and the RAND/VAR/SUP comparison design are from this package.

  5. supportsLabel-free channel ranks did not improve SEEG movement decoding over random selection sci:sha256:f9b6045c2f6fd997cfa5502b9cb132d0f398204e94a517a5de54bb2a542e48ed

    Same conclusion as the label-free kurtosis rank in the same features: no gain over random channels.

Cited by

0 publications

Public catalog entries on Sciscape.

No incoming citations on Sciscape.