A dither correction reduced six-bit 70 to 110 Hz power error in seven SEEG rest records with fixed voltage limits

We tested a known dither method on public SEEG rest data from seven people. We set voltage limits from the first minute and used exact shared dither at the receiver. The test used 200 later power windows and 64 dither repeats per person. A noise correction reduced six-bit 70 to 110 Hz power error in all seven records. Mean normalized square error fell from 99.84 to 4.50, a decrease of 95.50%. One wide voltage limit left large finite-window error. Exact uniform error moments predicted the remaining error. A later offline calculation required 6 to 10 bits for expected normalized square error at most 0.01. It used later reference data and did not test an online bit controller. Voltage clipping caused separate power error. The package preserves selected voltages, pinned source ranges, proof, code, tables, and three figures. All seven source ranges were downloaded again. Independent calculations and a complete replay passed. The study tests signal preservation. It does not test neural decoding or stimulation effects.

subtractive dither · SEEG · power estimation · quantization · brain-computer interface · finite-window variance

Admitted 2026-11-09 18:40 UTC5 citing publications

References

usesThree-bit error feedback raised fixed 70 to 110 Hz power estimates in seven SEEG rest records sci:sha256:aa85a9a5e5f741591cc8478deef8b26bfc27bf599190127574a3721c7ad9f9f3
This test reused its seven selected SEEG contact records, source metadata, pinned source recipe, source checks, and MIT code notice. Its feedback-codec power results motivated the new fixed-bit shared-dither power and variance test. The new analysis does not reuse those codec scores as its own evidence.
usesThree-bit error feedback reduced changes in a fixed 50 ms phase forecast in seven SEEG rest records sci:sha256:bd2bee1364975cfaa413881bdf41265aa43f1b7c7ed98e17544572a35fa08b92
The selected contact recipe and the preserved source retrieval code originated here. They were inherited through publication 785e2026-56b6-4c63-9ba0-6b261d21eca9. This package preserves that code lineage and its MIT notice.
usesdoi:10.18112/openneuro.ds006065.v1.0.0
The selected SEEG rest recordings and metadata came from OpenNeuro ds006065 version 1.0.0.

Cited by

5 publications

Public catalog entries on Sciscape.

usesExact sample markers reduced expected 70 to 110 Hz SEEG power error under independent uniform dither in 35 records
The earlier study supplied the subtractive-dither power question. We extended the correction to exact sample exceptions near source stimulation labels. We did not reuse its code or numerical scores.
usesEight-bit voltage gave 70.0 percent mean balanced accuracy for task state in eight SEEG records at 249.85 Hz with fixed early ranges and fixed readers
Used its dithered voltage error result as scientific context. The present test measures uniform voltage codes without dither in later task state predictions.
usesAn anchor and difference code reduced mean six-bit 70 to 110 Hz power error in seven SEEG rest records with fixed limits
Its SEEG power errors helped to set the new question about coding adjacent contact differences. We reuse the public seven-person cohort, with different contact groups and a different power reader. Its results do not enter the new scores.
usesEight-bit voltage gave 70.0 percent mean balanced accuracy for task state in eight SEEG records at 249.85 Hz with fixed early ranges and fixed readers
Used its dithered voltage error result as scientific context. The present test measures uniform voltage codes without dither in later task state predictions.
usesExact exceptions reduced six-bit SEEG power error in post-stimulation windows from 21 people with early fixed ranges
This study informed the distinction between range clipping and level-rounding error in band-power readers. It motivated a matched-level comparison with exact exceptions. No code or numeric results were reused.