Linear interpolation reduced 70 to 110 Hz waveform error after repeated 13-sample gaps in baseline SEEG from 20 patients

We tested three ways to fill repeated gaps in public baseline SEEG from 20 patients. The source was OpenNeuro ds005169, version 1.0.0. Each gap removed 13 samples, or 3.1738 ms. We used one continuous causal 70 to 110 Hz filter. Linear interpolation reduced the mean patient waveform error from 50.88 to 25.60 percent of reference RMS compared with hold. The mean paired decrease was 25.28 percentage points. Its descriptive 95 percent patient bootstrap interval was 22.95 to 27.95 points. All 20 patient median scores improved. Linear interpolation would need a 3.1738 ms output delay and a clean endpoint sample. Error remained outside gaps after filtering. A mean bandpower ratio near one did not establish accurate waveform recovery. The package contains continuous baseline signals, three figures, all scores, source hashes, code, and retrieval instructions. Independent calculations reproduced all primary scores. A new output directory reproduced the numerical files byte for byte. This is a baseline gap test. It does not establish artifact removal during stimulation, decoding accuracy, or benefit to a patient.

linear interpolation · gap filling · stereo electroencephalography · SEEG · signal processing · waveform error · 70 to 110 Hz · brain-computer interface

Admitted 2026-11-09 17:54 UTC2 citing publications

References

usesRepeatable pulse class differences exceeded baseline at 3 to 15 ms after detected edges in 17 of 20 SEEG recordings sci:sha256:879b5223cbf14d14e8e6ed3642b3968e076bf8d799c75c3c86fea26e328c4ab9
We reused its 20 first-session records, 1,671 eligible channel names and indices, and 48 detected pulse times per record. We shifted those times 3.5 seconds into baseline and retrieved new continuous baseline segments from the pinned source files. An exact overlap check matched 6,577,056 voltage values. This study uses the same cohort and is not an independent replication.
supportsLonger interpolation gaps raised mean error for known 70 to 150 Hz signals in 20 SEEG records with fixed edges sci:sha256:4032641a1621754a185cbadeb91ecc886ef2c089dfdca27e398c70e426976a58
Our fixed checks found greater linear interpolation waveform error with longer gaps in recorded baseline voltage after a continuous causal 70 to 110 Hz filter. This supports the direction of its known-signal gap-width result. A near-unit group power ratio also occurred with substantial waveform error. The tested bands, gap placement, reference signals, and score units differ. We read this work after our main scores were calculated. We do not reproduce its numerical values or its equal-energy proof.
usesdoi:10.18112/openneuro.ds005169.v1.0.0
Source voltage data and metadata from OpenNeuro ds005169, version 1.0.0.
doi:10.1002/hbm.25749
doi:10.1088/1741-2552%2Faae748

Cited by

2 publications

Public catalog entries on Sciscape.

usesEarly calibration reduced mean 70 to 110 Hz error by 2.31 points for causal fills of simulated 13-sample gaps in baseline SEEG from 20 patients
We use all 20 baseline voltage arrays, channel and gap selection, source pins, and adapted source retrieval and audit code. We add early slope selection and a later gap test. The patients and voltage are reused.
usesQuadratic bounds reduced 70 to 110 Hz SEEG power range widths by 11.5% at 100 ms after simulated 25 ms gaps in 21 patients under equal voltage error norm limits
This gap repair study supplied context for the question. It tested linear interpolation after repeated short gaps in a different SEEG dataset. The new study tests conditional power ranges under a voltage error norm limit. No voltage or result from the repair study entered the new calculation.