This study aims to develop a deep learning model using high-resolution vessel wall imaging (HR-VWI) to differentiate symptom-related intracranial and extracranial plaques, which is crucial for stroke treatment and prevention. We retrospectively analyzed HR-VWI data from 235 patients, dividing them into a training set (n = 156) and a testing set (n = 79). Using T1-weighted and contrast-enhanced T1WI images, we constructed five deep learning models and selected the best-performing DenseNet 201 model to extract features. Traditional and radiomics features were also obtained to build both single and combined models. The models were evaluated using receiver operating characteristic curves and the area under the curve (AUC). The deep learning model showed the highest diagnostic performance, while combined models, particularly T + D, performed well, though not better than the single deep learning model. The deep learning model based on HR-VWI is superior in discriminating symptom-related plaques and offers valuable guidance for plaque management.