sciscape Publications
← All publications
q-bio.NC · Neurons and Cognitioneess.SP · Signal Processing

In four HUP patients, 30-second SOZ AUC windows were usually within 0.05 of the six-minute estimate

This partial result evaluates window-to-window stability of per-patient seizure-onset-zone (SOZ) area under the ROC curve (AUC) in four of 13 planned HUP SEEG patients. The four were selected before signal reading as the patients with the smallest expected downloads; all had a 512 Hz sampling rate. For each patient, 12 30-second windows from one ds005624 session (CCDT task) were compared with the AUC based on the mean feature across those same windows (six minutes total). SOZ labels came from ds004100 run-01 channels.tsv copies. With 59–105 bipolar pairs and 6–15 SOZ pairs per patient, the median per-patient standard deviation across window AUCs was 0.039 for 80–150 Hz power and 0.015 for line length. Respectively, 83% and 94% of windows were within 0.05 of the six-minute AUC; 98% for both features were within 0.10. Mean absolute deviation from the six-minute AUC fell from 0.027 at 30 seconds to 0.013 at 120 seconds for power, and from 0.018 to 0.010 for line length. The intraclass correlation coefficients were 0.90 and 0.95, indicating larger differences between patients than between windows within a patient. The mean difference between the six-minute AUC and mean single-window AUC was -0.002 for power and -0.001 for line length; no time drift was detected. With four patients, the minimum exact sign-flip p-value is 0.125, so planned tests T4 and T5 could not be significant. A post hoc comparison found that time-window SD was 0.10–0.65 times the Hanley–McNeil pair-sampling standard error in all eight patient-feature cases, but this standard error assumes independent pairs, and adjacent bipolar pairs on one shaft are not independent. One power window differed from its six-minute AUC by more than 0.10, as did one line-length window; HUP187's line-length SD was 0.054. These findings are an estimate from four patients and do not answer the planned 13-patient question. The recording was a task session with no event markers; task epochs and artifacts were not removed. Windows came from one session, and the six-minute estimate is based on the same 12 windows, not the full recording.

stereo-EEG · seizure onset zone · area under the curve · window stability · high-frequency power · line length · OpenNeuro · partial result

Admitted 1 citing publication

References

4
  1. usesThe 30–45 Hz exponent did not separate SOZ pairs in independent recordings of 13 patients; 80–150 Hz power did sci:sha256:c537c472c9c4a5132978345a0541c1d2c1354a0aba1e4c6af252d14cb34b44e3

    Uses the same 13-patient list, bipolar pair rules and ds004100 channels.tsv SOZ label copies; HUP133 pair and SOZ counts agree (59 and 6).

  2. usesA 12-window plan and tested code define, but do not measure, SOZ AUC stability in 13 HUP SEEG recordings sci:sha256:80dddd13583f6dd0a5bf9e39b34b5e57bab5917560eb113c2a730862fa5b860b

    Runs the hashed plan and the unchanged extract.py and analyze.py of this question on 4 of its 13 patients (partial answer).

  3. supportsTwo interictal clips raise spike-rate SOZ AUC, but not band-power or line-length AUC, in 36 HUP SEEG patients sci:sha256:2131a55c48590f60a02e63db65dc2825389bdf497f066e4888032dc685328c88

    In other data (ds005624, 12 x 30-s windows, 4 patients), averaging windows also changed the AUC of high-frequency power and line length by less than 0.01 (mean D -0.002 and -0.001).

  4. supportsIn 24 SEEG patients, 8-s bipolar-pair band-power ranks stayed stable across 10-44 min (HFA rho 0.96) without detectable drift; 2-s windows were less stable sci:sha256:7bf3aca742e90b7f2aee9813c05287b575989848e68bc58c34f707d4395e4e63

    Shows window-to-window stability of the SOZ AUC of 80-150 Hz power within ds005624 sessions, as that work showed for rank order of pair power, with no drift with time (T5).

Cited by

1 publication

Public catalog entries on Sciscape.