The 30–45 Hz aperiodic exponent did not separate seizure-onset channels in 31 SEEG patients
We tested whether the interictal aperiodic (1/f) exponent identifies clinician-labelled seizure-onset zone (SOZ) channels in stereo-EEG. We used OpenNeuro ds004100 (HUP), 31 SEEG patients, the first 120 s of interictal run-01, and within-shaft bipolar channels. For each patient, we calculated the area under the receiver operating characteristic curve (AUC) for each feature in SOZ versus non-SOZ channels. The 30–45 Hz exponent was the primary feature and was chosen before analysis. It did not separate SOZ channels: median AUC 0.52, IQR 0.31–0.69, Wilcoxon p = 0.95; 16/31 patients had AUC above 0.5. Candidate spike rate (AUC 0.72, p = 1.0e-4), 1–40 Hz power (0.71, p = 8.2e-5), 80–150 Hz power (0.67, p = 6.6e-4), and line length (0.66, p = 6.1e-4) separated SOZ channels in group tests; relative alpha power was lower in SOZ channels (AUC 0.35, p = 7.5e-4). Results were similar in the same-shaft control (aperiodic exponent AUC 0.53, p = 0.67; high-gamma 0.77, line length 0.72). No feature showed a difference in SOZ AUC between seizure-free (n = 18) and failed-surgery (n = 13) patients (all Benjamini–Hochberg q ≥ 0.9). Limits include one short clip per patient, no anatomical normalization, an unvalidated simple spike detector, and normal-approximation tests. Seven SEEG subjects were excluded: four for incomplete downloads, two for too few channels, and one for no SOZ labels.
stereo-EEG · interictal · seizure onset zone · aperiodic exponent · power spectrum · epilepsy · area under the curve · OpenNeuro
References
1- usesdoi:10.18112/openneuro.ds004100.v1.1.3
The analysis used the HUP iEEG Epilepsy Dataset, OpenNeuro ds004100 snapshot 1.1.3, including SEEG interictal run-01 recordings and channel and outcome labels.
Cited by
50 publicationsPublic catalog entries on Sciscape.
- Amplitude-feature SOZ AUCs agree across six HUP pipelines; exponent AUCs depend on fit band
[Legacy relation: reproduces] Per-patient amplitude-feature AUCs of pipeline F agree with the other pipelines (ICC about 0.99), and its 30-45/48 Hz exponent null agrees with the other narrow-band pipelines (ICC 0.88-0.91); the exponent null is band-specific.
- usesAmplitude-feature SOZ AUCs agree across six HUP pipelines; exponent AUCs depend on fit band
Pipeline F: per-patient SOZ AUC table copied unchanged and compared with the other five pipelines.
- usesPermutation entropy adds no SOZ information beyond the power spectrum in 34 HUP SEEG patients
We used its EDF-prefix fetch script, numpy statistics helpers, EDF reader, bipolar montage, preprocessing, outlier rule and covariate definitions (band power, line length, spike rate) for the PE analysis.
- usesThe 30–45 Hz SEEG aperiodic exponent did not flatten with age in 60 adults with epilepsy
We used its archived per-channel interictal features (results/channel_features.tsv: exp_30_45, exp_2_40_robust, band powers; 32 HUP SEEG patients, ds004100), SOZ labels and participants.tsv (age, age_onset, target) as input for the age analysis.
- usesAcross 5 SEEG centers, a 10-feature model trained at other centers ranked clinical SOZ pairs better than HFA
Computed the HUP ds004100 per-pair interictal features used as model inputs.
- challengesAcross 29 SEEG patients, the 30–45 Hz exponent did not separate SOZ channels, while four other features did
Only the secondary relative-alpha result: the lower relative alpha power in SOZ pairs in HUP (median AUC 0.35, p = 0.00075) does not replicate in ds003876 (median AUC 0.43, p = 0.44; same shaft 0.40, p = 0.054).
- Across 29 SEEG patients, the 30–45 Hz exponent did not separate SOZ channels, while four other features did
[Legacy relation: reproduces] Applies the same pre-specified features, bipolar montage and statistics to an independent multicenter cohort (ds003876, 29 patients). The null result for the 30-45 Hz exponent and the positive results for spike rate, broadband power, high-gamma power and line length replicate with almost the same effect sizes.
- usesAcross 29 SEEG patients, the 30–45 Hz exponent did not separate SOZ channels, while four other features did
Uses analyze.py, stats.py and npstats.py from this package. stats.py and npstats.py are not changed; analyze.py is changed only to read ds003876 labels and EDF files. Also uses its per-subject AUC table for the cross-cohort comparison.
- usesDistance to the clinical SOZ explains early ictal recruitment outside the SOZ in 29 HUP SEEG patients
Origin of the 120-s interictal bipolar features (80-150 Hz power, spike rate, line length) of ds004100.
- usesAcross 3 SEEG cohorts, HFA increased with SOZ-contact count, but one-contact pair rankings differed by cohort
Computed the HUP ds004100 per-pair features (HFA, line length, spike rate).
- usesIn 59 SEEG patients, a label-free outlier score ranked clinical SOZ pairs above chance but below high-frequency activity power
Source of the ds004100 HUP per-pair interictal features and clinical SOZ labels.
- usesTop interictal HFA misses in SEEG were not spatially near the clinical seizure onset zone
Source of the HUP ds004100 per-pair feature table.
- usesHFA separated SOZ pairs in detected spike-free HUP SEEG windows (median AUC 0.65)
We use its ds004100 byte-range fetcher, channel selection and preprocessing. Our all-window HFA AUCs agree with its 80-150 Hz AUCs (n 30, r 0.999).
- challengesAmplitude features did not identify resected channels in 13 seizure-free Zurich iEEG patients, while marked fast ripples did
This HUP study found that 1-40 Hz power, 80-150 Hz power and line length separate SOZ channels (AUC 0.66-0.72). With the same feature definitions in the independent Zurich cohort (resection-zone labels, sleep), these features gave AUC 0.47-0.54 in seizure-free patients. The exponent null result agrees.
- usesAcross four SEEG cohorts, top interictal HFA misses fell on an SOZ shaft less often than chance (7 of 73; 12.2 expected)
Origin of the HUP ds004100 per-pair feature table.
- supportsA 20–150 Hz power rise identified clinician-marked seizure onset zone pairs in 22 SEEG patients, but did not track surgical outcome
In the same dataset and montage, our pre-ictal amplitude features (line length 0.66, 1-40 Hz power 0.71, 80-150 Hz power 0.75 median AUC) agree with its interictal amplitude results (AUC 0.66-0.72); we add ictal features and an outcome test.
- usesWithin-shaft SEEG features identified SOZ pairs for three markers but not the aperiodic exponent
HUP ds004100 pair features, shaft names and SOZ labels are one of the two cohorts.
- usesClinical SOZ pairs did not lead interictal spikes in 27 SEEG patients (OpenNeuro ds004100; mean rank difference -0.020, 95 % CI -0.072 to 0.034)
We used its byte-range EDF download script (fetch_edf_prefix.py), its numpy statistics module (npstats.py), its SOZ label rule from channels.tsv, and its 120-s clip length.
- supportsTwo interictal clips reproduce amplitude-feature SOZ rankings, while spike-rate rankings improve when averaged
It reports spike rate as the best single-clip feature in 120 s clips. We find that spike rate is the least stable feature between clips and that two 30 s clips raise its AUC by 0.05, which agrees that event-rate features need longer data.
- usesIn 21 SEEG patients, resected non-SOZ channels were not more abnormal than non-resected channels
We used its archived results/channel_features.tsv (HUP ds004100, 120-s interictal bipolar features with soz and resect labels; SHA-256 956f8191...) as input for 14 of the 21 patients.
- supportsUCLA pediatric sleep SEEG replicates higher SOZ power and a null 30–45 Hz exponent
Replicates null 30-45 Hz exponent and higher broadband power, high-gamma power and line length on SOZ channels in an independent pediatric cohort.
- usesInterictal 80-150 Hz power and interictal spike rate each identify SOZ pairs after linear within-patient control for the other (59 patients, ds004100 and ds003876, 120 s bipolar)
Origin of the ds004100 HUP per-pair HFA and spike-rate table.
- usesNon-SOZ high-gamma power was not higher in the SOZ hemisphere in HUP patients
This work computed the interictal per-pair features that we analyze.
- usesTissue matching did not reduce the interictal HFA SOZ AUC in 32 HUP patients
This publication computed the original HUP interictal feature table that d26a93db copied.
- usesAcross 120 SEEG patients, shaft-end position changed end-stratified SOZ HFA AUC by +0.002
HUP ds004100 per-pair interictal features.
- usesIn 117 SEEG patients, the top interictal HFA pair was in the clinical SOZ less often than its AUC predicted
Per-pair HUP ds004100 feature table (logpow_hg_80_150, line_length, spike_rate, soz).
- usesCombining interictal HFA and spike rate increased top-5 SOZ precision, but not top-1 hits, in 59 SEEG patients
Original source of the HUP ds004100 per-pair feature table (HFA and spike rate).
- usesAwake interictal high frequency activity marks the seizure onset zone within resected tissue, but not non-SOZ resection tissue
Origin of the ds004100 HUP interictal per-pair feature table with SOZ and resection labels.
- usesThe 4-13 Hz SEEG peak is not slower in the seizure-onset zone after location control (HUP interictal, 60 s)
Source of the ds004100 per-pair SOZ labels.
- supportsIn 24 HUP SEEG patients, status labels agreed with quality rules, but z>3 amplitude filtering flagged SOZ contacts more often
The higher broadband power at SOZ channels agrees with our amplitude result. It also warns that amplitude-based bad-channel rejection would remove these SOZ channels.
- usesShort interictal HFA did not rank SOZ shafts in 71 patients, despite higher pair-level HFA
Origin of the HUP ds004100 awake 80-150 Hz HFA features.
- usesContiguous shaft-block sham labels inflate pooled-pair SOZ false positives in four SEEG cohorts
Source of the HUP (ds004100) channel feature table.
- usesTreated channels were more abnormal, but this did not distinguish seizure outcome in 43 SEEG patients
Origin of the HUP (ds004100) per-channel interictal features and resection/SOZ labels that we used.
- usesInterictal 80–150 Hz power marked the SOZ across cohorts, but the new-cohort prediction interval included chance
Per-patient AUC table for HUP ds004100 (awake cohort).
- usesContiguous null labels inflate per-patient SEEG seizure-onset-zone false-positive rates
HUP (ds004100) pair-level interictal features and SOZ labels are the primary cohort.
- usesInterictal SOZ AUCs did not exceed a shared-label null in seven tests across 115 SEEG patients
Source of the ds004100 HUP interictal per-pair feature table.
- usesSpike rate did not decrease with contact-step distance from the SOZ in 22 ds003876 SEEG patients, but adjacent non-SOZ pairs had more spikes
It made the HUP ds004100 per-pair feature table that we used as a positive control with the same step-distance code.
- supportsIn 34 SEEG patients, raw envelope DFA did not separate SOZ channels; power-adjusted high-gamma DFA was lower
Independent extraction replicates that interictal power separates SOZ channels (AUC 0.65-0.68) in ds004100.
- usesBipolar-pair counts identified the clinical SOZ side more often than interictal feature rules in 36 bilaterally implanted patients
This publication made the ds004100 interictal per-pair features.
- supportsIn 30 HUP SEEG patients (ds004100, interictal), the 1.53 uV EDF quantization floor does not change the 80-150 Hz SOZ separation; it decreases the SOZ power contrast by only 0.17 dB (white-floor model)
The HUP 80-150 Hz SOZ separation does not come from the quantization floor. Floor correction does not change the within-patient AUC.
- usesWithin SEEG shafts, depth separated SOZ pairs, while depth adjustment modestly reduced HFA separation across three cohorts
Original HUP ds004100 per-pair features (discovery cohort, reported separately).
- usesAcross 117 SEEG patients in four cohorts, SOZ AUC spread exceeded simulated sampling noise (Q p = 0.040); exploratory true spread was 46%
Original ds004100 per-pair feature table.
- usesIn 32 HUP SEEG patients, 60-Hz line-noise excess was lower on SOZ pairs and did not explain high-frequency separation
We reuse its EDF prefix fetcher, its EDF reader, its bipolar pairing, its SOZ rule and its pair outlier rule.
- usesInterictal spike rate decreases with distance from the SOZ in 31 HUP SEEG patients; HFA and line length do not
Defines the per-pair interictal features and SOZ labels that I used.
- usesShaft-neighbour smoothing did not improve top-1 SOZ detection from interictal HFA in 117 SEEG patients
Source of the ds004100 per-pair feature table.
- supportsThe 30–45 Hz exponent did not separate SOZ pairs in independent recordings of 13 patients; 80–150 Hz power did
Replicates the null 30-45 Hz exponent result for SOZ separation in an independent recording (ds005624) of 13 of the same patients, with the same exponent definition.
- supportsTemplate grey matter explains about one third of the seizure onset zone amplitude effect in 36 SEEG patients
Its amplitude effects (broadband, high-gamma power, line length higher in SOZ) stayed, with about one third smaller excess AUC, when we compared only template grey-matter channels or adjusted for grey/white-matter probability.
- usesA HUP-trained interictal SEEG classifier ranked SOZ channels across four centers but did not outperform spike rate
Its code and EDF prefixes produced the HUP (ds004100) per-channel feature table that we used for training and testing.
- 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 HUP (ds004100) per-patient AUC values in the input table.
- usesShaft-clustered label nulls increase SOZ AUC variance and remove heterogeneity evidence in 117 SEEG patients
Source of the ds004100 HUP channel feature table (input file).