Seven connected difference channels preserved eight-contact SEEG decoder decisions without added error in eight public records
We tested whether seven differential channels can preserve a fixed eight-contact decoder in public SEEG data. We used OpenNeuro ds004473, version 1.0.1, with eight people. Three connected trees preserved all 1,128 later decisions when no coefficient error was added. Mean balanced accuracy stayed at 63.62%. Reconstruction used signed complex coefficients before power calculation. Under independent uniform errors added at edges, the star had 41.67% less expected squared error than the path. At the primary step of training coefficient RMS divided by 16, mean changed decisions were 0.91% and 1.28%. Errors added at contacts gave the same reconstruction across connected graphs. Four separate pairs changed 419 decisions. A small example showed why edge powers alone cannot recover centered contact powers. The package includes a transfer tool, proof, inputs, fixed models, scores, three figures, pinned source records, and repeat instructions. The data are reused from a prior publication. This test concerns an offline reading decoder. It does not test physical transmission, continuous intent detection, or brain stimulation.
stereoelectroencephalography · brain-computer interface · differential channels · signal reconstruction · graph Laplacian · ridge decoder · measurement error · common average reference
References
6- usesA median reference reduced mean accuracy loss from simulated white noise at 100 times the training scale on one unused contact in eight SEEG recordings sci:sha256:2df5339eddb5232562539c4891aea2a3649ae2cf867abfbafe410b930c1e9e02
We used its contact selection, trial audits, and fixed ridge method through the direct parent package.
- usesAn eight-contact common average reference changed fixed task decoder decisions in eight public SEEG records sci:sha256:7d0e58461450488051316af2a830bf23a0f9473798d3aedeea771c17feafc616
We used its unchanged Fourier inputs, eight-contact selections, trial splits, source records, and extraction code for new graph tests.
- usesSEEG power before a response gives 57.7% mean balanced accuracy for two cued keys on later trials in eight recordings sci:sha256:9fcbc6aad0f73578b656a03c940b9381ab5791e8f9875ee291d7fc423328ee2d
We used its source trial rules, metadata, and EDF methods through the retained upstream code.
- usesdoi:10.18112/openneuro.ds004473.v1.0.1
The analyzed records are from OpenNeuro ds004473 version 1.0.1; the archived source metadata and commit identify the dataset used.
- doi:10.1109/TSP.2007.912270
- doi:10.1088/1741-2552%2Facae0a
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesByte parity kept all six-bit SEEG decisions in eight offline records under one known packet loss and correct received bytes
Used all eight compact signal files, fixed reader models, graph definitions, and source check code. Extended the graph study with packed quantization, digital parity, and each known single packet loss.
- usesA contact covariance rule kept SEEG site decisions equal across equivalent bipolar trees in five offline patient tests
Its connected-difference transfer argument informed the matched-component scope and graph rejection checks. This study tests regularized site distances on different data.
- usesA loop check on 28 contact pairs missed gain faults in eight offline SEEG readers with no added output noise
We used its eight unchanged public SEEG Fourier files, eight fixed task reader models, trial audit records, and pinned source recipe. Its clean scores fixed our reader baseline.