What question did this study set out to answer?

To develop and validate a combined machine learning model integrating clinical factors, plaque characteristics, and perivascular adipose tissue (PVAT) radiomics from computed tomography angiography to predict short-term stroke risk in asymptomatic carotid stenosis.

January 21, 2026Frontiers in RadiologyOpen Access

Development of a machine learning-based radiomics model of perivascular adipose tissue for predicting stroke risk in patients with asymptomatic carotid stenosis: a multicenter study

Q: What does this research mean for the field?

A combined model integrating perivascular adipose tissue radiomics features from CTA with clinical factors effectively predicts short-term stroke risk in patients with asymptomatic carotid stenosis. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

View Full Paper

Ask AI

Bookmark

Why the study?

The study aimed to develop and evaluate a combined model integrating clinical features, conventional CTA features, and PVAT radiomics features to identify asymptomatic carotid stenosis patients at high risk for short-term stroke.

Does a combined model integrating perivascular adipose tissue radiomics and clinical features improve short-term stroke risk prediction in patients with asymptomatic carotid stenosis?

Key Points

To develop and validate a combined machine learning model integrating clinical factors, plaque characteristics, and perivascular adipose tissue (PVAT) radiomics from computed tomography angiography to predict short-term stroke risk in asymptomatic carotid stenosis.
Analyzed computed tomography angiography (CTA) imaging data from 582 patients across four cohorts, comprising a training cohort, an internal validation cohort, and two external validation cohorts.
Extracted radiomics features of carotid PVAT, selected nine optimal features, and evaluated five machine learning classifiers (KNN, SVM, LR, XGBoost, and BernoulliNB) to establish a radiomics score (Rad-score).
Identified independent clinical risk factors via univariate and multivariate logistic regression analyses to construct a combined clinical-radiomics prediction model and nomogram.
Hypertension (OR: 3.86, 95% CI: 1.97–7.82, p < 0.001) and soft plaque (OR: 4.29, 95% CI: 2.03–9.43, p < 0.001) were identified as independent clinical predictors of stroke.
Among the five classifiers, the XGBoost radiomics model achieved the best performance, with AUCs of 0.869 (95% CI: 0.808–0.930), 0.826 (95% CI: 0.731–0.921), 0.858 (95% CI: 0.787–0.929), and 0.848 (95% CI: 0.782–0.914) across the four datasets.
The combined model incorporating hypertension, soft plaque, and the Rad-score achieved AUCs of 0.911 (95% CI: 0.861–0.947), 0.868 (95% CI: 0.777–0.931), 0.882 (95% CI: 0.821–0.928), and 0.871 (95% CI: 0.808–0.920) across the four cohorts.

Structured PICO

Does a combined model integrating perivascular adipose tissue radiomics and clinical features improve short-term stroke risk prediction in patients with asymptomatic carotid stenosis?

Population

582 patients with asymptomatic carotid stenosis (ACS) confirmed by CTA, with no history of recent (within 6 months) ischemic stroke/TIA involving the ipsilateral carotid territory, and at least 2 years of complete follow-up data. Divided into training (n=188), internal validation (n=85), external validation 1 (n=157), and external validation 2 (n=152) cohorts.

Intervention

A combined prediction model integrating clinical factors (hypertension, soft plaque) and a radiomics score (Rad-score) derived from perivascular adipose tissue (PVAT) CTA images using an XGBoost machine learning classifier.

Comparator

Clinical model alone (incorporating hypertension and soft plaques) and other machine learning classifiers (KNN, SVM, LR, BernoulliNB).

Outcome

Short-term stroke occurrence (within at least 2 years of follow-up).hard clinical

A combined model integrating PVAT radiomics features and clinical factors effectively predicts short-term stroke risk in patients with asymptomatic carotid stenosis, offering superior discrimination compared to clinical factors alone.

Implication

Should not yet change practice in asymptomatic carotid stenosis; leaves open prospective validation of the radiomics-clinical model.

Abstract

Background: Our work aims to develop and evaluate a combined model that integrates clinical features, conventional computed tomography angiography (CTA) features, and radiomics features of perivascular adipose tissue (PVAT) to identify asymptomatic carotid stenosis (ACS) patients at high risk for short-term stroke. Methods: = 152). Radiomics features of PVAT were extracted from CTA images, and dimensionality reduction was performed to identify predictive features. Five machine learning classifiers were employed to construct radiomics models, and the model with the highest AUC was selected to generate the radiomics score (Rad-score). Clinical factors associated with stroke were determined using univariate and multivariate logistic regression analyses to construct a clinical model. A combined model integrating clinical factors and the Rad-score was subsequently developed, and a nomogram was created to provide a visual tool for stroke risk prediction. We assessed model performance comprehensively through calibration curves, discrimination analysis, reclassification, and clinical application. Results: A total of nine optimal radiomics features were ultimately selected from the CTA images. Across the four datasets, the AUC values of the five classifier models ranged from 0.643 to 0.869, 0.716 to 0.826, 0.651 to 0.858, and 0.638 to 0.848, respectively, with the XGBoost model demonstrating the best performance. The combined model, incorporating hypertension, soft plaque, and the Rad-score as variables, achieved AUCs of 0.911, 0.868, 0.882, and 0.871, respectively, across the four datasets. Conclusions: A combined model based on PVAT imaging features around carotid plaques can effectively predict the short-term stroke risk in ACS patients. This model may be expected to provide an important auxiliary tool for clinical prognosis assessment and treatment decisions, with potential clinical application value.