Repair norm budgets gave a median 21.5 ms guard at baseline RMS for a zero-phase 70–110 Hz filter in 400 SEEG pulses from five patients.
A filter can spread the effect of uncertain voltage near a stimulation pulse. We built an offline tool to bound this effect in a gamma readout. The tool uses the largest singular value of a finite filter operator. It searches for the first window start that meets an RMS limit in all tested later windows. We tested 400 public stereoelectroencephalography (SEEG) pulses from five patients in OpenNeuro ds004457, version 1.0.2. The main filter was zero-phase 70 to 110 Hz. The uncertain interval was −2 to 10 ms. Each output window had 62 samples at 2048 Hz. Starts were tested through 500 ms. Each input norm budget equalled the observed linear repair norm. Each RMS limit equalled the fixed baseline RMS. The median bound guard was 21.48 ms. The range was 10.25 to 105.47 ms. A 20 ms guard, rounded to the next sample, failed the bound in 225 of 400 pulses. The observed repair difference failed the same window rule in 27 pulses. All raw file hashes passed. Fresh extraction and a repeat analysis matched the preserved results. The input budget is an assumption. The method does not estimate true artifact energy or establish neural origin.
filtering · signal processing · SEEG · stimulation artifact · operator norm · RMS bound · gamma band · finite filters
References
4- usesBipolar subtraction changed site accuracy in both directions at 30–100 ms in five SEEG sessions sci:sha256:531e50239e932801e1c0cfd358414615addc29385833c8f1831bb538445b2800
The fixed contact choices and raw retrieval code came from this work through the parent package. Earlier decoder scores were not analysis inputs.
- usesReplacing pulse samples changed zero-phase 70–110 Hz SEEG voltage above baseline RMS at 20–50 ms in 27/400 pulses (five patients). sci:sha256:78299ccd5493ea3f4d9c8ec1dfad43bf77ec6ab49d467ba2f55a3b7b4806dddd
We reused 400 digital epochs, event selections, source hashes, and MEF extraction code. Fresh extraction matched all preserved arrays.
- usesdoi:10.18112/openneuro.ds004457.v1.0.2
The analysis uses SEEG recordings from this exact OpenNeuro dataset release; source hashes and retrieval instructions are preserved.
- doi:10.1523/JNEUROSCI.1325-22.2023
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesA voltage cap from the past second failed in 34 of 1104 simulated 10 ms gaps in 35 SEEG records
We used this related method for comparison in the report. It states a finite filter norm bound and the RMS reverse triangle inequality for offline signal repair. We read it after the new numeric analysis. It supplied no code, data, or numeric inputs to our calculations.
- usesA projected bound shortened 70 to 110 Hz power guards in 400 SEEG pulses under stated input error balls
This work supplied exact digital epochs, fixed contacts and events, retrieval and extraction code, and source metadata. We checked fresh raw files and wrote new power analysis code. Its filter results were not inputs to this study.