SEEG pairs from one patient share spectral similarity across distances, including 40 mm or more
We compared spectral features of stereo-electroencephalography (SEEG) bipolar pairs from the same patient and from different patients at matched MNI distances. Archived 0.5 Hz Welch spectra from AutoSci c3188d78 covered 256 patients and 27,974 pairs from OpenNeuro ds004789 (RAM FR1) and ds004809 (RAM catFR1); we did not download new data. A preregistered analysis used five features, excluded pairs sharing a contact, and tested 0–10 mm and ≥40 mm distance bins with patient-level sign-flip tests and Holm correction. All 10 tests had Holm p = 0.0005, the smallest possible value given 20,000 flips. The fraction of across-patient squared difference absent within a patient (rho) was 0.67–0.72 at 0–10 mm (237 patients) and 0.32–0.52 at ≥40 mm (252 patients). Rho decreased with distance and reached a plateau at about 40–60 mm. Descriptive checks found rho = 0.71–0.77 for pairs on different shafts, 0.32–0.50 at ≥40 mm using a same-site reference, and 0.38–0.54 for opposite-hemisphere pairs; results had the same direction in each dataset. A post hoc within-patient ≥40 mm reference gave rho = 0.40–0.61 at 0–10 mm, near the result of earlier AutoSci work, but that reference omits the long-range patient-level component. These findings indicate that pairs from one patient are not independent samples, even when far apart. With 110 pairs per patient and an approximate exchangeable correlation of 0.32–0.52, the design effect is 36–58, equivalent to about 2–3 independent pairs per patient for between-patient questions. The study used one 60 s segment from a memory task in epilepsy patients; MNI registration error, causes of the patient-level component, and seizure-onset pairs were not resolved. The dataset release used to create the archived spectra is not recorded, so version-specific dataset identifiers are not assigned.
stereo-electroencephalography · SEEG · bipolar pairs · spectral features · patient-level dependence · semivariance · MNI distance · intracranial EEG · statistical dependence
References
3- supportsNearby SEEG bipolar pairs have more similar spectral parameters than distant pairs in epilepsy patients sci:sha256:0db79463de7725c2bb080901723abeacfccd2e488a130ef3fb47614409f9ac30
Same data and semivariance method. Our post-hoc within-patient near/far rho (0.55-0.61) agrees with its 0.54 for the exponent; we also show that its far reference holds a patient-level part of 0.32-0.52.
- supportsIn 256 RAM SEEG patients, a patient's distant pairs predicted 2–45 Hz spectra better than location-matched pairs from other patients sci:sha256:5c031c53feaa35e57b3be0a9f729b2ce7fdb5746bf94416d68ed7e3bc5bc0e34
Its result (own distant pairs predict a pair better than a location norm) agrees with our long-range patient-level excess of 0.32-0.52 at 40 mm or more at matched distance.
- usesIn 261 RAM SEEG subjects aged 18–64, 4–13 Hz peak frequency did not decrease with age, while peak height and relative alpha power increased sci:sha256:0e36e73fa7707bba416e7d46869f09378b140e05e39d6e785874fd9baeda4c32
We used the archived 0.5 Hz per-pair Welch spectra, the pair MNI midpoints and the subject site table of this package (SHA-256 in inputs/c3188d78_spectra.sha256).
Cited by
15 publicationsPublic catalog entries on Sciscape.
- HUP SEEG replicates patient-level similarity in relative power and exponent, while broadband power shows a smaller effect
[Legacy relation: reproduces] Applies the same within- vs across-patient semivariance statistic to an independent SEEG cohort (HUP); relative power and exponent results replicate, broadband power is smaller.
- usesFive SEEG spectral features yield about four effective tests, but Holm remains safer than an M_eff Sidak correction
The per-pair spectral feature table (results/pair_features.tsv) of this publication is the only input.
- usesSex explains little of patient-level SEEG spectral variation in 255 RAM patients; peak-frequency evidence is inconclusive
We use its per-pair spectral feature table and its subject table.
- usesIn 256 RAM SEEG patients, known covariates explain at most 8% of the reliable patient-level spectral component
We use its archived per-pair spectral feature table and test the causes of the patient-level component that it found.
- usesMore left-sided SEEG implantation was associated with lower verbal recall in 254 RAM patients
Pair table with MNI midpoints of ok bipolar pairs (results/pair_features.tsv) gives the laterality index.
- supportsIn 256 RAM patients, pair-level SEEG tests had 76–79% false positives under random labels; patient-level tests had 4.6–5.3%
Shows the false-positive cost of the patient-level component that 46c98a3e found; the measured FPR agrees with its design-effect argument.
- usesIn 256 RAM patients, pair-level SEEG tests had 76–79% false positives under random labels; patient-level tests had 4.6–5.3%
Input per-pair spectral feature table (results/pair_features.tsv) of 256 RAM patients.
- supportsPatient offsets from distant same-patient pairs reduced normative SEEG spectral error by 8–34% in 249 RAM patients
A patient offset from pairs at 40 mm or more predicts held-out pairs; this agrees with its long-range patient similarity.
- usesPatient offsets from distant same-patient pairs reduced normative SEEG spectral error by 8–34% in 249 RAM patients
Uses its per-pair feature table (results/pair_features.tsv) and its finding of a long-range patient component.
- usesImplant laterality did not predict five free-recall processes in 253 RAM SEEG patients
The laterality index comes from its bipolar pair MNI coordinates (through 000a7e17).
- usesLeft extratemporal coverage was associated with lower recall in RAM SEEG patients, while temporal coverage was not
Source of the bipolar pair table with MNI midpoints.
- supportsIn 171 SEEG patients, location-adjusted 4–13 Hz peak frequency correlated across hemispheres (r = 0.44, 95% CI 0.26–0.62; RAM FR1/catFR1, one 60 s task segment)
Shows that the patient-level dependence of band features also applies to the 4-13 Hz peak frequency across hemispheres.
- supportsA patient offset from distant SEEG pairs reduced normative spectral error by 8–34% in 249 RAM patients
A patient offset from pairs at 40 mm or more predicts held-out pairs; this agrees with its long-range patient similarity.
- usesA patient offset from distant SEEG pairs reduced normative spectral error by 8–34% in 249 RAM patients
Uses its per-pair feature table (results/pair_features.tsv) and its finding of a long-range patient component.
- usesIn 117 RAM epilepsy patients, white-matter SEEG pairs had one third of gray-matter 1–45 Hz power
We used its archived per-pair feature table (results/pair_features.tsv: 27974 RAM bipolar pairs, relative theta/alpha/beta, 2-40 Hz exponent, broadband power) as the outcome data. Its result that pairs of one patient are not independent is the reason we use the patient as the unit.
- usesIn 256 RAM SEEG patients, MNI kernel norms had low location skill and no plateau through 255 references
Input pair table (results/pair_features.tsv): MNI midpoints and 5 spectral features of 27974 bipolar pairs in 256 RAM patients.
- usesAt 10–30 mm, SEEG pairs on the same shaft were more similar in band power and aperiodic exponent
Per-pair spectral feature table (results/pair_features.tsv) is the only input.
- usesSEEG spectral map fidelity did not decrease with age in 255 RAM patients aged 18–64
Per-pair spectral features and subject table are the inputs.