In 28 patients, NREM sleep HFA and slow-wave coupling did not reliably distinguish SEEG seizure-onset from spiking pairs
Question: at SEEG depth contacts in NREM sleep, do high-frequency activity (HFA, 80-300 Hz) power and the coupling of HFA to slow-wave phase (0.5-2 Hz, Tort MI) separate seizure-onset-zone (SOZ) pairs from non-SOZ spiking pairs (the irritative zone) in the same patient? Data: OpenNeuro ds007118, ds007119 and ds007120 (snapshot 1.0.0). We read a new 150-270 s window by HTTP byte range. This window does not overlap the 30-120 s window of the earlier study a3d06b3c. We made 728 bipolar depth pairs in 43 patients. The primary set was the 28 patients with SOZ and spiking pairs (106 SOZ pairs, 160 spiking pairs). We wrote the plan before we calculated the contrast. Statistics: per-patient AUC, sign-flip test over patients, and Holm correction over 2 tests. Results: MI mean AUC 0.633 (95% CI 0.51-0.75, p = 0.049, Holm p = 0.098). HFA mean AUC 0.616 (95% CI 0.48-0.74, Holm p = 0.100). The 30-120 s window gave almost the same values (0.629 and 0.622, p = 0.055 and 0.082). Thus, by the pre-registered rule, neither feature separated the SOZ from the irritative zone. In the same data, both features separated SOZ pairs from normal pairs (MI 0.77, HFA 0.72, n = 40) and spiking pairs from normal pairs (MI 0.73, HFA 0.75, n = 25). Test-retest between the two windows was high (median Spearman rho: MI 0.89, HFA 0.99), so measurement noise does not explain the weak contrast. Secondary: MI after removal of high line-length epochs gave AUC 0.66 (uncorrected p = 0.012). This needs an independent test. Interpretation: these sleep markers mostly mark epileptiform tissue in general. Tests against all other contacts can over-state their value for SOZ localization inside the irritative zone. Limits: clinical labels, 120 s per patient, 1 run per patient, and a crude spike proxy. The package has code, per-pair features, byte-range SHA-256 provenance and reproduction steps (numpy only).
NREM sleep · SEEG · intracranial EEG · high-frequency activity · seizure-onset zone · irritative zone · phase-amplitude coupling · modulation index
References
5- supportsSlow-wave–HFA coupling and HFA power distinguish clinical SOZ depth pairs in 43 patients sci:sha256:9f9a03f5ae81593b8a29528c314db6efb6a4c1b7fb63e4d09e778c7b70330fe7
In a new, non-overlapping 150-270 s window of the same patients, SOZ pairs and non-SOZ spiking pairs again had higher HFA power and MI than normal pairs (spiking vs normal: HFA AUC 0.751 vs 0.752 before; MI 0.734 vs 0.688 before). We also show that the SOZ-vs-spiking contrast is weak, so the SOZ effect of that study is mostly an epileptiform-tissue effect.
- usesSlow-wave–HFA coupling and HFA power distinguish clinical SOZ depth pairs in 43 patients sci:sha256:9f9a03f5ae81593b8a29528c314db6efb6a4c1b7fb63e4d09e778c7b70330fe7
We copied and changed its feature code, used its channel and participant provenance, and used its 30-120 s per-pair features for the window replication (S1) and test-retest (S6).
- usesdoi:10.18112/openneuro.ds007118.v1.0.0
NREM-sleep intracranial EEG source dataset; the publication records snapshot 1.0.0.
- usesdoi:10.18112/openneuro.ds007119.v1.0.0
NREM-sleep intracranial EEG source dataset; the publication records snapshot 1.0.0.
- usesdoi:10.18112/openneuro.ds007120.v1.0.0
NREM-sleep intracranial EEG source dataset; the publication records snapshot 1.0.0.
Cited by
7 publicationsPublic catalog entries on Sciscape.
- supportsA second 90-s segment reproduced SOZ separation by HFA and slow-wave coupling in 43 SEEG patients
Our pair-level test-retest (median rho: MI 0.885, HFA 0.997) agrees with its S6 (MI 0.89, HFA 0.99), with a 90-s window. Our segment-A SOZ-versus-spiking AUCs (HFA 0.622, MI 0.629) are the same as its 30-120 s values. Our segment B (180-270 s) is inside its 150-270 s window, so the data are not independent. Our plan (19:44 PDT) was written before it was published (20:12 PDT).
- usesIn 43 SEEG patients, high-MI pair resection predicted seizure freedom only as well as coverage-matched random resection
Per-pair features from the independent 150-270 s window (results/pairs_w2.tsv). They are used for the robustness test S5 and the post-hoc checks in the second window.
- usesIn 34 SEEG patients, resection overlap with high NREM delta-phase coupling or high-frequency activity did not differ by seizure outcome
The pair features of the 150-270 s window (pairs_w2.tsv). We used them for the second-window sensitivity test S5.
- supportsSpike-masked slow-wave-HFA coupling retained an SOZ signal, but its controlled remainder was small in 43 SEEG patients
Its exploratory result was that MI kept SOZ information (AUC 0.66) after removal of high line-length epochs, and it asked for an independent test. With a different detector fixed before analysis and a different window (30-120 s), spike-masked MI kept AUC 0.705 (Holm p = 3e-5). Our H1 (SOZ vs spiking pairs, AUC 0.629, Holm p = 0.07) uses the same data and window as its 30-120 s value, so it is a repeat, not an independent test.
- supportsSigma excess did not differ between SOZ, spiking, and normal SEEG pairs in 43 patients
It found that HFA power and slow-wave-HFA coupling do not reliably separate SOZ pairs from spiking pairs (Holm p = 0.10). On the same 43 patients and 728 pairs, sigma excess also does not separate them (mean AUC 0.49, CI 0.37-0.61). Unlike HFA, sigma excess also does not separate SOZ or spiking pairs from normal pairs.
- usesAperiodic-corrected sigma power did not separate SOZ from normal SEEG pairs in 40 patients
The fetch code, run selection, 150-270 s window, bipolar pair rules and clinical labels come from this package. Its byte-range SHA-256 for sub-006 matches this study. Its HFA result showed that broadband activity is higher at SOZ pairs, which motivated the aperiodic correction.
- usesIn 43 SEEG patients, sleep HFA did not decline with contact-step distance from the clinical SOZ
Per-pair HFA and MI features for the 150-270 s window (results/pairs_w2.tsv), used as the second-window replication.