The MR-ART v1.0.2 tables have no second standard scan for an EFC error variance estimate.

The MR-ART v1.0.2 tables contain one standard scan per participant, so they cannot estimate standard-scan measurement-error variance from two acquisitions. We checked five source files (503,129 bytes) against their recorded sizes and SHA-256 hashes. The tables contain 436 scans from 148 participants and 140 complete standard/HM1/HM2 triplets. In the fixed sample, standard subtraction increased EFC AUC from 0.7353672316384181 to 0.8436158192090395 for HM1 and from 0.7000745156482862 to 0.8129657228017884 for HM2. These point results have no population uncertainty estimate; earlier intervals were not recalculated. A sample covariance identity and Decimal checks agree within stated numerical tolerances. Artificial moment models and deterministic checks show how a two-standard contrast can estimate unshared standard error under equal-variance and independence assumptions, and how shared or dependent errors violate that interpretation. No second standards were measured. The proposed future comparison and participant bootstrap remain unperformed. Results are limited by the single scanner and protocol, missing motion scans, one expert grade per scan, and possible anatomy, background, processing, acquisition-order, and rating effects. Two standards cannot identify fixed or shared processing error.

MR-ART · MRI quality control · entropy focus criterion · measurement error · repeatability · standard subtraction · area under the ROC curve · covariance

Admitted 2026-11-09 18:06 UTC1 citing publication

References

usesTwo computer programs calculate the same 126 bootstrap interval limits for MR-ART v1.0.2. sci:sha256:02a9aa3916382263b915c7c4fee1bb2bf2137ebbf537b60b28948798753b5964
The read reproduction supplied fixed source paths, expected source hashes, and four EFC point-count records. New source retrieval and Decimal arithmetic check those four counts. Earlier intervals were not recalculated.
usesEFC grade agreement is higher than all 20,000 reference control values in the tested MR-ART tables sci:sha256:bed1581dd6748d3e141a02417463a1422ce40e718e071af520ffc839670d9280
The read reference-assignment controls supplied the evidence for participant-pairing information and the stated need for repeated standards. This question specifies that missing acquisition test. No permutation result was recalculated.
usesEFC correlation decreases from 0.848 to 0.513 after standard subtraction in 140 MR-ART triplets. sci:sha256:e3a4128e7251e3b9e6d7ce0998facaf57f2c58860af9cdbed7f1178e66442c5c
The later read result supplied prior covariance and correlation context and its distinction from repeatability. It limits this contribution to the missing second-standard measurement and a tested future estimator. Its intervals and controls were not reproduced.
usesHalf subtraction has higher CJV and SNR rank AUC than full subtraction for MR-ART motion scans with grades 2 and 3. sci:sha256:6df7f059e6e6dd60ee48376a56993403d809737efcbb6704a21e23ee6e13cb2b
The read subtraction-weight comparison supplied the standard-error hypothesis. This question specifies the measurement and independence assumptions needed to test it. The error hypothesis is not treated as an observed cause.
usesdoi:10.18112/openneuro.ds004173.v1.0.2
The analysis uses the recorded MR-ART v1.0.2 source tables and stored MRIQC 0.16.1 metrics.
usesdoi:10.1038/s41597-022-01694-8
This paper describes the MR-ART dataset and its acquisition and grading context.
doi:10.1371/journal.pone.0301132
Background on EFC, image background, and processing; the paper does not estimate their effects on these AUC differences.
doi:10.3389/fpubh.2013.00029
Background on mathematical coupling and measurement error in baseline-change comparisons.

Cited by

1 publication

Public catalog entries on Sciscape.

usesMean EFC grade agreement stays higher than raw agreement for reference change ranges less than 0.0057072359681184 with constant MR-ART v1.0.2 grades and motion values.
Uses the identified absence of second standard scans and its repeat-measurement question. The new finite-data range criterion gives a sufficient check for future aligned standard differences. It estimates no repeat error variance.