Contrast-enhanced dual-energy CT angiography-derived features used in combination with additional parameters to develop a reliable multidimensional risk prediction model for acute stroke

This study addresses the pressing need for more accurate stroke risk prediction, as acute stroke remains one of the leading causes of mortality and disability worldwide. The authors developed a multidimensional prediction model that integrates carotid plaque characteristics, vascular lumen parameters, perivascular adipose tissue (PVAT) measurements obtained from contrast-enhanced dual-energy CT angiography (DE-CTA), and serum lipid biomarkers. The retrospective analysis included 212 patients from two centers, divided into a training cohort and an external validation cohort. Patients were categorized as symptomatic, if MRI showed ipsilateral acute anterior circulation infarcts, and as asymptomatic otherwise. Using univariate analysis and LASSO regression, the team constructed a multivariate logistic regression model. Model performance was assessed with multiple statistical tests, showing robust predictive accuracy in the external validation cohort, with an area under the curve of 0.810, high sensitivity and moderate specificity. These findings suggest that combining contrast-enhanced DE-CTA-derived plaque and PVAT features with lipid biomarkers provides a reliable and clinically useful approach to predicting acute stroke risk, potentially enabling earlier screening and intervention.