Application of abdominal CT radiomic features for discriminating prevalent coronary heart disease in patients with type 2 diabetes mellitus
5 affiliations
- 1Department of Radiology, China-Japan Union Hospital of Jilin University
- 2Department of Endocrinology and Metabolic Diseases, China-Japan Union Hospital of Jilin University
- 3Weifang People’s Hospital
- 4Department of Radiology, The Second Hospital of Jilin University
- 5Department of Anesthesiology, China-Japan Union Hospital of Jilin University
Abstract
Purpose: To develop and validate a hybrid model combining pancreatic-aortic radiomic features from non-contrast abdominal CT with clinical factors for discriminating prevalent coronary heart disease (CHD) status in patients with type 2 diabetes mellitus (T2DM).
Materials and methods: This retrospective study enrolled 407 T2DM patients (277 with CHD, 130 without), randomly split into training (n = 284) and internal test (n = 123) sets. Pancreatic auto-segmentation used a 3D V-Net, while the abdominal aortic wall was manually segmented by blinded radiologists. Following Image Biomarker Standardization Initiative (IBSI)-compliant preprocessing, radiomic features underwent reproducibility filtering, collinearity removal, LASSO regression and coefficient threshold filtering (|β|>0.05). Five models were built using logistic regression, support vector machine and Gaussian process classifiers. Model performance was assessed by AUC, calibration plots and decision-curve analysis (DCA).
Results: Seven pancreatic, nine aortic and three combined radiomic features were retained. The hybrid model yielded a numerically higher test-set AUC of 0.84 (95% CI: 0.77–0.91). Its discriminative performance was numerically greater than the clinical model (AUC = 0.65, 95% CI: 0.56–0.74; P < 0.001), pancreatic radiomic model (AUC = 0.70), aortic radiomic model (AUC = 0.82), and combined radiomic model (AUC = 0.83). It exhibited favorable calibration, with a net clinical benefit comparable to or numerically greater than the alternative models across threshold probabilities of 10%–90% on DCA.
Conclusion: The hybrid clinical-radiomic model improves discrimination of prevalent CHD status in T2DM patients, supporting opportunistic non-invasive identification of existing CHD using routine non-contrast abdominal CT.
Related
- CT radiomic features detect mineral, collagen, and water contributions to cortical bone texture
- Evaluation of MRI Radiomic Features as Surrogates of Tumor Immune Status in Breast Cancer: A Pilot Study of MRI-Derived Radiomic Features for Non-Invasive Immune Profiling in Breast Cancer.
- Hybrid deep learning and meta learning for Alzheimer disease classification via multimodal radiomic and visual features
- Effect of Intensity Discretization on Radiomic Features in En Face EDI-OCT of Choroidal Nevi
- Lobar analysis of classical and deep learning radiomic features in chronic airflow limitation
BibTeX
@article{sui2026application,
title = {{Application of abdominal CT radiomic features for discriminating prevalent coronary heart disease in patients with type 2 diabetes mellitus}},
author = {He Sui and Yingchun Wan and Ming Qu and Yun Zhang and Xu Sun and Tao Li and Lin Liu and Zhanhao Mo},
journal = {Frontiers in Endocrinology},
year = {2026},
month = nov,
doi = {10.3389/fendo.2026.2006373},
url = {https://www.frontiersin.org/articles/10.3389/fendo.2026.2006373/full},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Application of abdominal CT radiomic features for discriminating prevalent coronary heart disease in patients with type 2 diabetes mellitus
AU - He Sui
AU - Yingchun Wan
AU - Ming Qu
AU - Yun Zhang
AU - Xu Sun
AU - Tao Li
AU - Lin Liu
AU - Zhanhao Mo
PY - 2026
DA - 2026/11/16
JO - Frontiers in Endocrinology
DO - 10.3389/fendo.2026.2006373
UR - https://www.frontiersin.org/articles/10.3389/fendo.2026.2006373/full
AB - Purpose: To develop and validate a hybrid model combining pancreatic-aortic radiomic features from non-contrast abdominal CT with clinical factors for discriminating prevalent coronary heart disease (CHD) status in patients with type 2 diabetes mellitus (T2DM). Materials and methods: This retrospective study enrolled 407 T2DM patients (277 with CHD, 130 without), randomly split into training (n = 284) and internal test (n = 123) sets. Pancreatic auto-segmentation used a 3D V-Net, while the abdominal aortic wall was manually segmented by blinded radiologists. Following Image Biomarker Standardization Initiative (IBSI)-compliant preprocessing, radiomic features underwent reproducibility filtering, collinearity removal, LASSO regression and coefficient threshold filtering (|β|>0.05). Five models were built using logistic regression, support vector machine and Gaussian process classifiers. Model performance was assessed by AUC, calibration plots and decision-curve analysis (DCA). Results: Seven pancreatic, nine aortic and three combined radiomic features were retained. The hybrid model yielded a numerically higher test-set AUC of 0.84 (95% CI: 0.77–0.91). Its discriminative performance was numerically greater than the clinical model (AUC = 0.65, 95% CI: 0.56–0.74; P < 0.001), pancreatic radiomic model (AUC = 0.70), aortic radiomic model (AUC = 0.82), and combined radiomic model (AUC = 0.83). It exhibited favorable calibration, with a net clinical benefit comparable to or numerically greater than the alternative models across threshold probabilities of 10%–90% on DCA. Conclusion: The hybrid clinical-radiomic model improves discrimination of prevalent CHD status in T2DM patients, supporting opportunistic non-invasive identification of existing CHD using routine non-contrast abdominal CT.
SN - 1664-2392
ER -