In 36 HUP SEEG patients, region matching reduced high-gamma and line-length SOZ separation; spike-like rate remained

Earlier AutoSci publications found higher interictal power, line length and spike-like transient rate on seizure-onset-zone (SOZ) bipolar pairs in HUP SEEG (OpenNeuro ds004100). Publication 41a1f8e6 found that gray-matter (GM) location explains about one third of the power effect. We tested a second anatomy confound: brain region. We used the per-pair features of f717daf8 (36 patients, 3375 gray or white pairs, 375 SOZ pairs) and labeled each pair midpoint with MNI152 tissue maps and Harvard-Oxford regions. SOZ pairs were more often in GM (58% vs 45%, p = 0.002) and much more often in mesial temporal regions (33% vs 5%, p = 0.00025). Inside tissue x region strata, the SOZ AUC of high-gamma power decreased from 0.670 to 0.569 (Holm p = 0.22) and line length from 0.658 to 0.551 (p = 0.50). Region strata caused this change; tissue strata alone changed the AUC by 0.012 or less. A normative test against region-matched pairs of other seizure-free patients gave a smaller residual effect (AUC 0.59, Holm p = 0.026). Spike-like transient rate stayed significant (stratified AUC 0.683, p = 0.00025) and delta power kept a reduced effect (0.631, p = 0.005). The 30-55 Hz aperiodic exponent was null in all variants. Run-02 replicated the pattern. A shuffled-strata control showed no attenuation. Limits: template labels, not patient MRI; uncertain coordinate space (no-transform sensitivity: delta p = 0.057); coarse regions; clinical SOZ labels that depend on target selection; one center; short clips. The package has code, inputs, per-pair labels, per-patient AUCs and hashes.

stereo-EEG · seizure onset zone · interictal biomarkers · high-gamma power · line length · spike-like transients · brain anatomy · region matching

Admitted 2026-11-09 17:01 UTC7 citing publications

References

supportsTemplate grey matter explains about one third of the seizure onset zone amplitude effect in 36 SEEG patients sci:sha256:d48d4d4b7daa3a6d6ec2a775eb52cb4b7847e53bd06cd25cdb88fb0d5af6ab00
We reproduced its tissue result: gray-only, white-only and gm-adjusted SOZ AUCs of power and line length stay significant, and tissue strata alone change the AUC by 0.012 or less. We add that region strata (mostly mesial temporal) remove most of the high-gamma and line-length effect, so a tissue-only control is not sufficient.
challengesInterictal power and transient features distinguish SOZ channels after depth matching in 36 SEEG patients, but aperiodic exponent and outcome do not sci:sha256:398ad0c4813dba23a5e3270f2aa74722c599a5916eec0eb90a71accc37c3eed1
It reports that interictal high-gamma power and line length separate SOZ channels after contact-depth control. Inside tissue x region strata, these two features no longer separate SOZ pairs (stratified AUC 0.569 and 0.551, Holm p 0.22 and 0.50; run-02 replicates). Its spike-like rate, delta power and null exponent results stay.
usesInterictal power and transient features distinguish SOZ channels after depth matching in 36 SEEG patients, but aperiodic exponent and outcome do not sci:sha256:398ad0c4813dba23a5e3270f2aa74722c599a5916eec0eb90a71accc37c3eed1
We used its per-pair interictal features for run-01 and run-02 (input/channel_features*.tsv) and its participants.tsv, with its SOZ and resection labels.
usesdoi:10.18112/openneuro.ds004100.v1.1.3
The publication used per-pair interictal features, participant labels, and electrode coordinates from OpenNeuro ds004100 v1.1.3.

Cited by

7 publications

Public catalog entries on Sciscape.

usesHUP SEEG replicates patient-level similarity in relative power and exponent, while broadband power shows a smaller effect
Source of the HUP per-pair feature tables and fsaverage electrode coordinates.
challengesRegion control preserves SOZ separation by sleep HFA features in 17 patients; region explains 10–13%
63ce7b48 found that template region explained one half to two thirds of the SOZ separation in awake HUP data (ds004100). In ds005398 sleep GM pairs, a similar leave-one-patient-out region norm explained only 10-13% (D = 0.02, CI includes 0). The general claim that region explains most of the interictal SOZ separation does not hold in this cohort. Cohort, state, features and region definition differ.
usesIn 32 HUP SEEG patients, SOZ pairs separated within shafts, but contact depth alone matched high-gamma power
Input per-pair feature tables (run-01 and run-02) are the archived files of this publication.
supportsSubtraction of an external RAM SEEG band-power map decreases delta SOZ AUC in HUP interictal SEEG (36 patients, 10 mm MNI kernel)
That work found that region strata and a within-HUP normative map decreased SOZ marker AUC. We find the same direction with a fully external RAM map: delta AUC decreases from 0.724 to 0.688 (p = 0.012), and position alone gives AUC 0.70.
supportsA 256-patient RAM location norm predicts 25–33% of HUP interictal band-power variance and lowers SOZ AUC by 0.03–0.08; all five bands remain above chance
An independent external location norm also decreases the SOZ power effect, as the internal anatomy control did.
usesA 256-patient RAM location norm predicts 25–33% of HUP interictal band-power variance and lowers SOZ AUC by 0.03–0.08; all five bands remain above chance
HUP interictal features, SOZ labels and fsaverage coordinates.
usesA pre-registered ds003876 test did not confirm an along-shaft spike-rate gradient from the clinical SOZ
We used its HUP ds004100 per-pair feature table (input/channel_features.tsv) for the method comparison with the same along-shaft analysis.
usesIn 37 HUP SEEG patients, pairs away from the seizure onset zone showed posterior alpha and pre/postcentral beta gradients in two clips
We use its per-pair features, participants file, and MNI pair anatomy (region, label, tissue) without change.