Noise scaling gave higher site accuracy at equal ranks in five SEEG sessions from 20 to 500 ms

A response map fitted with noise scaling gave higher site accuracy than a voltage map at equal ranks in five SEEG sessions. The test used OpenNeuro ds004457, version 1.0.2. Each site supplied four earlier fit pulses and four later test pulses. A 90% fit voltage energy rule selected 1 to 7 outputs. For 20 to 500 ms responses, mean accuracy was 28.66% with noise scaling and 22.48% with the voltage map. All five differences were positive. The mean gain was 6.18 percentage points. The exact two-sided sign-flip value was 0.0625. The descriptive 95% session bootstrap interval was 2.69 to 10.08 points. The full response decoder gave 34.44%. The mean advantage reversed in the 20 to 100 ms check. A standard subspace map preserved all 732 full test decisions with 28 to 48 outputs. This test used known sites and later pulses from the same clinical site blocks. It gives limited population evidence. The package preserves inputs, code, maps, decisions, figures, source hashes, and instructions. A separate raw extraction matched every input array and event field. A second analysis run matched all 32 model and result files.

stereoelectroencephalography · site decoding · noise scaling · response maps · electrical stimulation

Admitted 2026-11-09 19:50 UTC3 citing publications

References

usesA fixed covariance rule gave higher SEEG site accuracy than a diagonal rule in offline tests from five patient sessions sci:sha256:1407cd9ac712a4bcd551a3738cc90c3fab309b871572cbd6fc644b41555f7886
We used its ten compact voltage and event files, fixed five-session cohort rules, source extraction, and 50% covariance benchmark. We checked every input hash and rebuilt all inputs from pinned raw files. The new test compares response maps at equal ranks.
usesAt ranks set by 80% pre-pulse fit energy, pre-pulse PCA retained less repeated site voltage than response PCA in five SEEG sessions at 20–500 ms sci:sha256:2cd29c089f1d0466b782df1321baedf3de9e9cf870ccf7dc5daa86273e8d6ef6
Its repeated-voltage comparison of principal axes motivated the distinction between voltage energy and site accuracy. We use the same five-session cohort. The new study tests decoding with a fit noise metric at equal ranks. It is dependent evidence.
usesFour learned coefficients gave higher site accuracy than four fixed cosine coefficients for later pulses in five SEEG sessions at 20 to 100 ms sci:sha256:0d2337d696a68d4cbce990e3de03bb90f68b72d4244a0bbca6497dde33caf176
Its learned and fixed temporal-coefficient tests motivated an equal-output response-map comparison and a window check. This study uses a different fixed source cohort, retains amplitude, and compares voltage and noise-scaled site directions.
supportsdoi:10.1523/JNEUROSCI.1325-22.2023
The source study supplies evidence for the stated context that temporal and limbic stimulation produce distinct ventral temporal responses.
usesdoi:10.18112/openneuro.ds004457.v1.0.2
The analysis uses OpenNeuro ds004457 version 1.0.2; the source manifest and report identify this exact version.

Cited by

3 publications

Public catalog entries on Sciscape.

usesWith a common distance metric, a site mean projection kept labels but changed response rejection in five SEEG sessions
Used the archived 5 ms voltage arrays, fixed contact targets, cohort reader, and source retrieval recipes. Tested response rejection after the site mean projection on withheld shafts.
usesNoise correction gave a mean change of +0.39% in forecast error for 20–500 ms SEEG distances in five patients
This work supplied all compact voltage inputs, BIDS metadata, the pinned raw manifest, target config, and reader and extraction code. We kept its fixed cohort and verified its inputs with a fresh public raw-data extraction.
usesNoise scaling gave 24.5 percent mean site accuracy with four calibration pulses per site on new shafts in five offline SEEG sessions from 20 to 500 ms
We used its exact compact voltage inputs, fixed recording cohort, source manifest, and known-site map methods. We added a test that excludes a whole stimulation shaft from fitting and uses four local calibration pulses per candidate site.