An analysis of abdominal CT scans found that measuring the texture of total abdominal fat improved detection of coronary calcification beyond measuring fat area alone. The texture-based approach also had accuracy comparable to a more complex model combining the best-performing radiomic features from subcutaneous and visceral fat.
The study aimed to characterize the CT profiles of subcutaneous and visceral abdominal fat, identify sex-specific and pro-atherogenic signatures, and assess whether those features added value for cardiovascular risk assessment.
What the scans measured
The analysis included 1,736 adults without known coronary artery disease. Researchers used non-contrast CT to extract size, first-order and second-order texture features from subcutaneous abdominal fat, visceral abdominal fat and total abdominal fat, referred to as SAF, VAF and TAF. Here, radiomics means turning image patterns into measurable features, rather than relying only on the area of fat.
The team used partial least squares discriminant analysis and XGBoost classification models to distinguish the abdominal-fat depots, biological sex and coronary calcification. Calcification was defined as a coronary calcium score greater than zero.
Fat depots carried different signals
SAF and VAF had distinct radiomic profiles. SAF showed lower attenuation, or lower measured density on the scan, and greater structural continuity in high-density regions.
Sex-related differences appeared in both depots and were especially pronounced in SAF. Male scans showed greater textural heterogeneity, meaning more variation in the image pattern, and greater radiodensity than female scans.
Textural heterogeneity also marked the study’s pro-atherogenic signatures, the patterns examined in relation to atherosclerosis-related risk. Both SAF and VAF showed increased heterogeneity in these signatures.
For coronary calcification, the density signal ran in opposite directions across the depots. Calcification was associated with higher SAF radiodensity but lower VAF radiodensity. The authors describe these patterns as metabolic densification and pathological hypertrophy, respectively.
The model result, in context
Taken together, the authors interpret texture heterogeneity as a key feature of male, visceral and pro-atherogenic adipose-tissue phenotypes. They report support for protective characteristics of SAF relative to VAF.
The authors present total abdominal fat radiomics as a potential alternative to assessing the two depots separately for individualized obesity risk stratification. In the model comparison, the total-fat approach improved performance over TAF area and had accuracy comparable to more complex combined-feature models.
The comparison is notable because the more complex models combined the best-performing pro-atherogenic features from SAF and VAF, while the TAF approach used radiomic features from the total abdominal-fat compartment.
These findings are associations, and they do not show that a particular fat pattern causes coronary calcification. A better-performing imaging model also does not by itself show that TAF radiomics improves clinical decisions or cardiovascular outcomes.
The supplied analysis gives the direction of the associations and the model comparison, but not detailed effect estimates, model metrics, uncertainty estimates or external validation. That leaves open whether the findings would reproduce across populations, scanners and CT acquisition protocols, and whether TAF radiomics would add clinically meaningful value beyond fat-area measurements.
Publication and data
The article was accepted on 31 July 2026 and published as the version of record on 20 August 2026. Its funding statement reports support for JM from Fundação Amélia de Mello through the Bolsa de investigação CUF & D. Manuel de Mello as part of SAFE-CT. The authors declared no competing interests.
The study data are available through the cited Zenodo record with restrictions and from the authors on reasonable request, with permission from the Faculty of Medicine of the University of Porto. The funding statement identifies the SAFE-CT study with registry identifier NCT06438393.
Paper data and sources
Original title: Sexual dimorphism and pro-atherogenic computed tomography radiomic phenotypes of subcutaneous and visceral abdominal fat.
Authors: Alice Ribeiro Lopes, Inês Sousa, Fabio Nunes et al.
Journal/Repository: International journal of obesity (2005)
Status: Peer-reviewed
First online: 2026-08-20
DOI: 10.1038/s41366-026-02195-z
Original paper · Full text