Half subtraction has higher CJV and SNR rank AUC than full subtraction for MR-ART motion scans with grades 2 and 3.
This fixed sensitivity analysis compares raw MRIQC scores with subtraction of 0.5 or 1 times each participant's standard-scan reference in OpenNeuro MR-ART ds004173 v1.0.2. The grade-rank comparison includes 278 motion scans from 147 participants, with grades 2 and 3, across HM1 and HM2; these conditions use five and ten head-nod instructions and do not measure head movement. The two conditions have equal weight in the mean. Half subtraction gives higher mean rank AUC than full subtraction for CJV (0.803759 vs 0.768333) and total SNR (0.840003 vs 0.783334). For EFC, half subtraction is higher than raw (0.785530 vs 0.717721) and lower than full subtraction (0.828291). The 95% pointwise participant-bootstrap interval for half minus full is [0.011973, 0.061423] for CJV, [-0.110832, 0.024260] for EFC, and [0.022721, 0.092644] for total SNR. The interval for EFC and the intervals for half minus raw CJV and SNR include zero. These intervals have no multiple-comparison adjustment. The analysis does not establish equivalent performance, an optimized weight, or a clinical rule. It uses one scanner and protocol, grades 2 and 3 only, and data used in earlier metric selection; there is no external test sample. The supplied metrics were not recalculated from images.
MR-ART · MRI · image quality · motion artefacts · reference subtraction · rank AUC · bootstrap · CJV · EFC · signal-to-noise ratio
References
4- Standard-reference subtraction improves EFC but reduces CJV condition-matched grade ranking in MR-ART sci:sha256:5246205abcf30a204e7b86cba2a1fa9d96a7077d08902ee630dba181fa75a49d
[Legacy relation: reproduces] Decimal arithmetic at precision 60 digits gives the same 12 condition-specific raw and full subtraction point counts. This relation covers those counts only. It does not cover earlier bootstrap intervals or image processing.
- usesStandard-reference subtraction improves EFC but reduces CJV condition-matched grade ranking in MR-ART sci:sha256:5246205abcf30a204e7b86cba2a1fa9d96a7077d08902ee630dba181fa75a49d
This result supplied the grade-2/3 rank comparison, metric directions, source paths, expected checksums, and reason for the specified half-weight comparison. This run downloaded all five source files again and wrote new calculation code.
- usesdoi:10.18112/openneuro.ds004173.v1.0.2
The analysis uses the five supplied metric and grade files from this exact OpenNeuro version. Saved metadata records CC0 and the matching version-specific DatasetDOI.
- doi:10.1038/s41597-022-01694-8
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesThe MR-ART v1.0.2 tables have no second standard scan for an EFC error variance estimate.
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.
- usesStandard subtraction with calculated coefficients increases EFC pair agreement in the tested MR-ART tables.
This result uses the earlier comparison with constant coefficients 0, 0.5, and 1. They give the reference methods for the calculated coefficients.