OBJECTIVE: This study is aimed at exploring the value of a high-resolution vessel wall imaging (HR-VWI)-based radiomics approach for identifying symptomatic basilar atherosclerotic plaques. MATERIALS AND METHODS: This retrospective study included 154 patients, hospitalized at Shanxi Cardiovascular Hospital for cerebrovascular disease, who had basilar artery stenosis and underwent HR-VWI examinations between July 2020 and December 2023. The patients were randomly divided into training and validation sets in a 7:3 ratio and further classified into symptomatic and asymptomatic groups based on the presence or absence of acute infarction lesions in the basilar artery supply area, respectively. Clinical and traditional imaging features were collected for all patients. Manual plaque segmentation and radiomics feature extraction were performed on the layers of basilar artery plaques identified in the plain-scanning and enhanced HR-VWI images. Feature denoising was performed using minimum redundancy maximum correlation, and the least absolute shrinkage and selection operator (LASSO) algorithm was applied to reduce feature variables. The most suitable features were selected through univariate and multivariate logistic regression analyses. The plain-scanning, enhanced, and plain - scanning + enhanced models were established. Finally, a combined model was constructed by integrating selected clinical and traditional imaging features (intraplaque hemorrhage [IPH], enhanced plaque signal) with the plain - scanning + enhanced HR - VWI radiomics features. The recognition efficiency of each model for symptomatic basilar atherosclerotic plaques was evaluated using the receiver operating characteristic (ROC) curve and the area under the receiver operating characteristic curve (AUC). RESULTS: A statistically significant difference in age was observed between symptomatic and asymptomatic groups (p < 0.05). Conventional imaging features, including IPH, vessel diameter at stenosis, lumen area at stenosis, enhanced plaque signal, stenosis rate, and plaque load, also demonstrated statistically significant differences between the two groups (p < 0.05). Firth-penalized logistic regression analysis identified IPH and enhanced plaque signal as risk factors for symptomatic basilar artery plaques. In the training set, the AUC values for detecting symptomatic plaques were 0.813, 0.853, and 0.898 for plain-scanning, enhanced, and plain - scanning + enhanced models, respectively. In the validation set, the AUC values were 0.786, 0.848, and 0.880, respectively. The efficacy of models was in the order of plain - scanning + enhanced model > enhanced model > plain - scanning model. The combined model, integrating IPH and enhanced plaque signal with the plain - scanning + enhanced HR - VWI radiomics features, achieved an AUC of 0.916 in the training set, with sensitivity, specificity, and accuracy of 0.821, 0.941, and 0.879, respectively, for predicting symptomatic basilar artery plaques. In the validation set, the AUC value was 0.891, with sensitivity, specificity, and accuracy of 0.792, 0.870, and 0.830, respectively. The combined model demonstrated superior performance compared with the plain - scanning + enhanced model. CONCLUSION: Our HR-VWI-based radiomics models can accurately distinguish symptomatic from asymptomatic basilar atherosclerotic plaques. It is superior to the traditional model in the identification of high-risk plaques.