Drift terms at original scan times blocked error transfer from excluded scans after cosine projection in 431 fixed AOMIC v2.0.0 task designs under full rank least squares
We tested whether errors in excluded fMRI scans can change a task contrast after a fixed cosine projection. We used 431 complete AOMIC v2.0.0 working memory designs. The response and design received the same filter. Scans with framewise displacement greater than 0.5 mm and scan zero were then removed. Of 431 runs, 162 had motion-flagged scans. Errors on those scans had nonzero contrast influence in all 162 runs when the final model had no added drift terms. Adding the original cosine terms at original scan times blocked transfer. Terms built on renumbered kept scans did not block it. For an error with RMS one on flagged scans, the first model had a median bound of 0.009931 and a maximum of 1.081266 reference standard deviations. The reference assumes independent original response errors with variance one. A proof shows equality to a direct joint fit on kept scans when inputs are fixed and the design has full rank. All 12,930 fits passed the rank check. Independent solutions, error witnesses, source hash checks, and a repeat run passed. These are fixed design sensitivity bounds. No measured brain bias or neural effect is estimated. The package contains a NumPy audit tool, compact public inputs, source records, saved operators, results, figures, and repeat instructions.
fMRI · motion censoring · cosine projection · contrast influence · least squares · error bounds · design audit
References
4- usesResponse-only removal of six motion terms reduced known unit contrasts in 431 fixed AOMIC v2.0.0 working memory designs after shared drift removal sci:sha256:061d9a1371b95e2dddc49d3124d269c10d26416b55b4f8728ff06b3f8fa82087
The earlier test of response-only motion removal informed the question about processing order. The present test applies the same filter to response and design, then tests scan removal and drift timing. No code or results from that earlier test are reused.
- usesBoth cohorts had 95% intervals that included zero for mean extra log contrast variance from motion mask timing in 431 AOMIC v2.0.0 working memory designs with fixed white noise sci:sha256:044463f720d58b5a648cb98667f1d2d085da23c7a7e3a169ead69bb00da47a91
Reused the 431 complete event and motion records, image metadata, pinned source plans and hashes, and unchanged task builder and retrieval scripts. They made the public design audit and fresh source checks possible.
- usesdoi:10.18112/openneuro.ds002785.v2.0.0
The study used event, motion, and image metadata from AOMIC PIOP1 version 2.0.0 to construct and audit 207 fixed task designs.
- usesdoi:10.18112/openneuro.ds002790.v2.0.0
The study used event, motion, and image metadata from AOMIC PIOP2 version 2.0.0 to construct and audit 224 fixed task designs.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- supportsThe same nuisance span recovered retained GLS contrasts after subtraction in 431 full rank AOMIC working memory designs with fixed AR(1) rho 0.4
The earlier audit showed that the original cosine nuisance span blocks influence from removed scans under full rank OLS. We support that established mechanism and extend its tested scope to joint motion and drift subtraction, recovery from cleaned retained data, and fixed-covariance GLS. We did not reuse its code or numeric results.
- usesFull-run cosine projection biased cross-part products in 431 fixed AOMIC task designs under independent scan noise
We used all 431 complete AOMIC event, motion, and image metadata records, the pinned source plans and hashes, and the unchanged task builder and retrieval scripts. We applied its original-time drift projection identity to each scan part. This enabled the new cross-part covariance audit and the local-fit control.