AI-assisted whole-pancreas MRI measurements correlate with CT attenuation
Source check: Source checked. Not independent scientific confirmation.
Finding: In 158 paired MRI/CT examinations, whole-pancreas fat fraction and CT attenuation showed a Spearman correlation of −.755. A locally trained U-Net assisted part of the MRI annotation, with manual correction of every assisted segmentation. The retrospective study also found substantial slice-to-slice variability; it did not independently validate automated segmentation or disease prediction.
Technology: U-Net pancreas annotation assistance and CNN-based liver segmentation
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- A U-Net trained during the annotation process assisted later MRI masks before manual correction; a separate CNN estimated liver fat. Learned segmentation aided measurement collection, while conventional correlations provided the scientific result.
- Test and result
- Of 543 potentially paired examinations, selection/exclusion yielded 158 subjects with MRI and non-contrast CT within six months, plus 59 MRI-only examinations for 217 total MRI measurements. Mean pancreatic MRI fat fraction correlated inversely with CT attenuation (Spearman −.755); slice-to-slice MRI variation increased with mean whole-pancreas fat fraction (Spearman .51). The prior CT-only study is acknowledged separately.
- Limitation / still unknown
- All pancreas masks were manually drawn or manually corrected after CNN assistance. The source does not quantify how many final subjects required learned assistance, and this is not a model-accuracy validation. The locally trained MRI U-Net requires independent testing according to the authors. Referral-based inclusion, exclusion of inflammatory/neoplastic disease, image-quality selection, imperfect MRI resolution and up-to-six-month modality separation constrain interpretation. No pancreatic histological ground truth or equivalence of fat measures is established by correlation, and demographics/diabetes associations are observational. The source extends earlier CT-only measurements; it is not a first discovery that fat and CT attenuation are related.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC12314592/
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: linked separately.
Review date: 21 Nov 2026. Review scope: Original publication metadata, computational/annotation Methods, Results and bounded prior-work/limitations Discussion passages actually read..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 1/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.