Why the study?
This study aimed to develop a left ventricular myocardium-based radiomics signature from CCTA to predict functional coronary artery stenosis in a relatively large sample of patients with suspected or known stable CAD.
Population
3480 patients with suspected or known stable CAD undergoing CCTA
Comparison
Clinical-CAD-RADS-radiomics model vs CAD-RADS-radiomics model and LVM-based radiomics signature
Design
Prediction model development and validation study
Answers are grounded in this paper and cite their sources.
Radiomics adds no incremental value for ischemia prediction beyond CAD-RADS; leaves open utility in prospective or randomized settings.
We aimed to develop left ventricular myocardium (LVM)-based radiomics signature from CCTA for predicting functional coronary artery stenosis in patients with suspected or known stable coronary artery disease (CAD) with a relatively large sample size. A total of 3480 patients were enrolled. The patients from November 2018 to December 2019 ( n = 2584) were divided into training set and testing set with a ratio of 7:3, respectively. Then, the patients between January 2020 and December 2020 ( n = 896) were included into the validation set. The functional coronary artery stenosis was defined as computed tomography–derived fractional flow reserve (CT-FFR) ≤ 0.80. Univariate and multivariable logistic regression analysis was performed to analysis the association between the presence of functional coronary artery stenosis and radscore, clinical risk factors and coronary computed tomography angiography (CCTA) features. The models were evaluated using calibration, receiver operating characteristic (ROC), and decision curve analyses. Finally, 13 features were used to construct the radiomics model and the radiomics model exhibited good discrimination ability in predicting myocardial ischemia and the AUC values was 0.745 in training cohort, 0.744 in test cohort and 0.733 in validation cohort, respectively. The clinical-CAD-RADS-radiomics exhibited under the curves (AUCs) of 0.932, 0.936 and 0.915 for predicting myocardial ischemia in the training, test and validation cohort and revealed minimal improvement of predictive performance than CAD-RADS-radiomics model in the training ( p = 0.072), test ( p = 0.021) and validation ( p = 0.158) cohort, respectively. The LVM-based radiomics signature was associated with an increased risk of myocardial ischemia. However, no incremental prognostic benefit of LVM-based radiomics signature over CCTA features was detected for myocardial ischemia in patients with suspected or known CAD. Clinical trial number: Not applicable.
No takes yet. Share an insight, caveat, or question.
Huang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: