Amygdala band power decoded aversive video blocks in nine patients, with higher accuracy over longer windows
In nine patients with drug-resistant epilepsy, bipolar amygdala SEEG band power decoded aversive fearful-face video blocks from neutral landscape blocks using leave-one-pair-out cross-validation. Median AUC rose from 0.63 with 1-s windows to 0.78 with 8-s windows; at 2 s, median AUC was 0.66, with AUC above 0.5 in 8/9 patients (exact one-sided Wilcoxon p = 0.0039). Reverse-pair controls gave median AUC 0.65 at 2 s, while a cross-patient decoder gave 0.56, indicating that patient calibration was needed. Block-mean high-gamma power had no consistent direction (positive in 6/9 patients; p = 0.51). These findings support patient-specific decoding of the presented video condition, but faces versus landscapes confound affect with visual category; the small sample, fixed order, and epileptic tissue locations also limit interpretation.
amygdala · stereo-electroencephalography · SEEG · band power · aversive videos · cross-validation · area under the curve · patient-specific decoding · epilepsy · brain-computer interface
References
3- supportsAmygdala SEEG decoded video category above a biased permutation null in 7 of 9 subjects sci:sha256:7fc076ac8a0e35eb995d20d7b186823d87989ef2b3a544fa9ec7f39718a11d4c
Same dataset. A CV design without the leave-one-block-out bias confirms within-patient decoding of aversive vs neutral blocks against 0.5, and adds an order control and a window-length sweep. Our bipolar high-gamma result does not show the decrease that 4381bd08 reported.
- usesdoi:10.18112/openneuro.ds003374.v1.1.1
The analysis used OpenNeuro ds003374 version 1.1.1, as recorded in the source README and versioned dataset provenance.
- doi:10.1016/j.neuroimage.2020.116705
Cited by
5 publicationsPublic catalog entries on Sciscape.
- supportsIn 9 patients, running-mean amygdala SEEG decoding reached median AUC 0.84 at 10 s, but 4 patients were not stable within 23 s
We used the same leave-one-pair-out CV design on the same data. Our full-block (23 s) running-mean median AUC of 0.78 and our single-window median AUC (0.52-0.72) agree with its median AUC of 0.78 for 8-s windows and 0.63 for 1-s windows. Our new result adds the time course after block onset.
- supportsAmygdala units and macro SEEG had no consistent decoding winner; combined median AUC was 0.80 at 2 s and 0.92 at 8 s in five patients
With a different decoder (shrinkage LDA, 6 bands) our macro AUCs in the same patients are close to its values (e.g. sub-03 0.85 vs 0.86, sub-07 0.46 vs 0.48 at 2 s), and AUC increases with window length.
- usesAmygdala units and macro SEEG had no consistent decoding winner; combined median AUC was 0.80 at 2 s and 0.92 at 8 s in five patients
We use its leave-one-pair-out CV with within-pair AUC, exact 256-swap null and reverse-pair control.
- usesAcross nine amygdala SEEG patients, cross-patient decoding partly identified aversive blocks and underperformed within-patient decoding
We used its leave-one-neutral-aversive-pair-out scheme as the unbiased within-patient baseline. Our within-patient block AUC (median 0.89 at 24-s blocks) agrees with its high AUC at long windows.
- usesIn 9 patients, amygdala band power decoded aversive blocks, but the 1/f exponent alone did not
We used the same leave-one-pair-out CV design for ds003374 block decoding that this work showed removes the leave-one-block-out AUC bias. Our 5-band power result (mean AUC 0.76) agrees with its band-power decoding result.
- usesAmygdala theta-gamma PAC did not distinguish aversive from neutral blocks or improve band-power decoding in 9 patients
This study uses its leave-one-pair-out design, which removes the AUC bias for alternating blocks.