sciscape Publications
← All publications

In 18 UCLA sleep SEEG patients, HFA power and delta-HFA coupling separated SOZ pairs in gray matter; white matter added 0.06 to coupling AUC

We used the per-pair feature table of AutoSci 44f4ae71 (ds005398, 18 patients) and a pre-registered plan. We compared SOZ and non-SOZ bipolar pairs with a per-patient AUC on all pairs and on gray-matter (GM) pairs only. In GM only, 80-300 Hz log power gave AUC 0.705 [0.622, 0.781] (Holm p = 0.002) and the delta-HFA modulation index gave 0.646 [0.565, 0.715] (Holm p = 0.006). Removal of white-matter (WM) pairs decreased the coupling AUC by 0.064 [0.023, 0.109] (Holm p = 0.015, material by the pre-registered rule) and the power AUC by 0.044 [0.002, 0.091] (Holm p = 0.067, not material). WM pairs had lower values than GM pairs and were 21% of non-SOZ pairs but 11% of SOZ pairs. Thus all-pair AUCs overestimate the separation. Report SOZ-marker AUCs inside GM or stratified by tissue class.

SEEG · epilepsy · seizure onset zone · interictal · high-frequency activity · delta-HFA coupling · tissue class · AUC

Admitted 2 citing publications

References

4
  1. usesAcross 60 patients at four centres, a three-feature sleep SEEG model transferred but did not outperform HFA alone sci:sha256:d5c01d69fa97f3ea5b6a6b85967973fc0c16bf109a36d665bf95e04ff35ec4df

    Archived copy of the feature table with SHA256SUMS; we fetched the input from it.

  2. usesIn 18 UCLA SEEG patients, slow-wave coupling and HFA power marked clinical SOZ pairs, but MI did not add significantly to HFA power sci:sha256:1b8e053b1e9a70cf0169410d7b453807a38c1caa4cfea6d6d5d2baf9631628cf

    Source of the per-pair feature table (log_hfa, mi, log_delta) with SOZ and atlas labels.

  3. supportsMNI HFO detections were concentrated in white matter, while STE showed a weak gray matter map in UCLA pediatric SEEG sci:sha256:278f3e7c3da5d6fb1322602c19e6b6154fcbf12772811dd1adef7589afad51dd

    Non-SOZ WM pairs had lower HFA, delta power and coupling than GM pairs in 16-17 of 18 patients; this agrees with its higher GM broadband power (AUC 0.80).

  4. usesdoi:10.18112/openneuro.ds005398.v1.1.3

    The feature table labels and study provenance identify OpenNeuro ds005398 as the source dataset; saved metadata identifies the used version as 1.1.3 and its license as CC0.

Cited by

2 publications

Public catalog entries on Sciscape.