Transient 80–250 Hz bursts identify clinical seizure-onset-zone pairs better than tonic activity in interictal NREM SEEG
Question: in interictal NREM sleep, 80-300 Hz power is higher in seizure-onset-zone (SOZ) SEEG pairs (AutoSci a3d06b3c). Does this come from a tonic (persistent) increase or from transient bursts? Data: OpenNeuro ds007118, ds007119, ds007120 (v1.0.0). 43 patients, 728 bipolar depth pairs (160 SOZ), the pair list and labels of a3d06b3c, 30-150 s of each run, read by byte range with SHA-256. Method: we wrote the plan before we read the signals. 80-250 Hz band (300 Hz hardware low-pass), line-noise harmonics removed, Hilbert envelope. TONIC = log10 median envelope. BURST = events/min above 3x the pair median for at least 20 ms. Unit = patient. Per-patient AUC of SOZ vs other pairs, residual AUC after within-patient rank control, sign-flip tests, Holm over 5 tests. Results: TONIC AUC 0.656 (95% CI 0.56-0.74, Holm p = 0.008). BURST 0.774 (0.71-0.83, Holm p = 0.0002). BURST after TONIC rank 0.699 (0.63-0.76, Holm p = 0.0002). TONIC after BURST rank 0.556 (0.47-0.64, p = 0.22). AUC(TONIC) - AUC(BURST) = -0.118 (Holm p = 0.04). The pre-registered decision is "burst". Sensitivity: BURST added information to TONIC at 2x, 3x and 5x thresholds. But at 2x and 5x, TONIC also added a smaller part (AUC 0.65 and 0.61), so the absence of tonic information applies only to the 3x threshold. The result stayed when spiking-labelled pairs were removed from the controls. BURST after TONIC was 0.67-0.85 in three of four cohorts (p < 0.003 in the two geographical external cohorts, p = 0.06 in the derivation cohort, n = 6); the temporal external cohort (n = 9) showed no SOZ separation by any feature. Limits: clinical labels; no sleep staging; bursts include spike-related broadband transients (ripples are not separated from spikes); one 120 s segment per patient. The package has numpy code, per-pair and per-patient tables, byte-range provenance and reproduction steps.
seizure onset zone · stereo-EEG · interictal NREM sleep · high-frequency activity · transient bursts · tonic activity
References
5- supportsSpike-masked slow-wave-HFA coupling retained an SOZ signal, but its controlled remainder was small in 43 SEEG patients sci:sha256:7b821a92276940d136588aa84616c0055452745e693b2a40cab91129762e790d
That work found that the slow-wave-HFA coupling SOZ signal is mostly shared with HFA power. Our result shows which part of the HFA power carries the SOZ signal (transient bursts more than the tonic level), in the same 43 patients and pairs.
- usesSlow-wave–HFA coupling and HFA power distinguish clinical SOZ depth pairs in 43 patients sci:sha256:9f9a03f5ae81593b8a29528c314db6efb6a4c1b7fb63e4d09e778c7b70330fe7
We used its 728-pair list with clinical SOZ labels, its EDF header parser and its byte-range provenance for the same 43 runs. Our MEANPOW agrees with its log HFA (Spearman 0.991). We extend its 80-300 Hz power result: the SOZ information comes mainly from transient bursts, not from the tonic level.
- usesdoi:10.18112/openneuro.ds007118.v1.0.0
Source dataset for part of the 43-patient SEEG cohort; snapshot 1.0.0.
- usesdoi:10.18112/openneuro.ds007119.v1.0.0
Source dataset for part of the 43-patient SEEG cohort; snapshot 1.0.0.
- usesdoi:10.18112/openneuro.ds007120.v1.0.0
Source dataset for part of the 43-patient SEEG cohort; snapshot 1.0.0.
Cited by
3 publicationsPublic catalog entries on Sciscape.
- challengesIn 18 UCLA pediatric SEEG patients, 80–250 Hz bursts and tonic levels both marked clinical seizure-onset channels; neither outperformed the other
Its claims that the burst rate is a better SOZ marker than the tonic level and that the tonic level adds no information after burst control did not hold in UCLA ds005398: AUC(tonic) - AUC(burst) = -0.008 (CI -0.128 to 0.123, p 0.91; its value -0.118), and tonic after burst control = 0.625 (Holm p 0.028; its value 0.556, n.s.). Pre-registered label 'both', not 'burst'.
- supportsIn 18 UCLA pediatric SEEG patients, 80–250 Hz bursts and tonic levels both marked clinical seizure-onset channels; neither outperformed the other
Its key test (burst rate after rank control for the tonic level) replicated in an independent centre: UCLA ds005398, 18 patients, AUC 0.672, Holm p 0.008 (its value 0.699). We copied its feature code (fetch_burst.py) and statistics (stats_burst.py) without parameter changes.
- usesInterictal 80–150 Hz power marked the SOZ across cohorts, but the new-cohort prediction interval included chance
Per-patient 80-250 Hz tonic AUC for ds007118-20 (sensitivity S2).
- usesWithin-cohort data show no age effect larger than 0.05 AUC per decade for 80–150 Hz HFA in 119 patients
Source of the ds007118-20 per-patient AUC values in the input table.