The abstract is read at the publisher; the summary is JClub's.
Automated CT-derived biomarkers of pancreatic perivascular adipose tissue (PVAT) distinguish individuals with type 2 diabetes mellitus (T2DM) from those without, achieving AUC 0.93 with specificity 0.91.
- Why it matters: Intrapancreatic fat is established for T2DM, but pancreatic PVAT has been understudied, leaving uncertainty about whether its imaging signal can identify current disease and future risk.
- What they did: A fully automated 3D deep neural network segmented pancreatic anatomy and surrounding structures on contrast-enhanced CT, then derived intrapancreatic, extrapancreatic, and PVAT biomarkers; logistic regression with feature selection built T2DM classifiers.
- The result: PVAT biomarkers differed significantly between diabetes and non-diabetes groups, and the best multivariate model reached AUC 0.93; exploratory analysis for incident diabetes within 4 years showed similarly high discrimination (AUC 0.93).
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.