Multimodal Radiomics Model Combining HR-VWI and Clinical Features for Identifying Symptomatic Basilar Atherosclerotic...

Area of Science:

• Cardiovascular Imaging
• Radiology
• Artificial Intelligence in Medicine

Background:

• Basilar artery atherosclerotic plaques pose a significant risk for cerebrovascular events.
• Accurate identification of symptomatic plaques is crucial for timely intervention and stroke prevention.

Purpose of the Study:

• To evaluate the efficacy of a high-resolution vessel wall imaging (HR-VWI)-based radiomics approach in distinguishing symptomatic from asymptomatic basilar atherosclerotic plaques.
• To compare the performance of radiomics models with traditional imaging features.

Main Methods:

• A retrospective study of 154 patients with basilar artery stenosis undergoing HR-VWI.
• Radiomics features were extracted from plain-scanning and enhanced HR-VWI images.
• Machine learning algorithms (LASSO, logistic regression) were used for feature selection and model development, including a combined model with clinical features (intraplaque hemorrhage [IPH], enhanced plaque signal).

Main Results:

• Radiomics models showed increasing efficacy with plain-scanning + enhanced HR-VWI.
• The combined model achieved an Area Under the Curve (AUC) of 0.916 in the training set and 0.891 in the validation set for identifying symptomatic plaques.
• IPH and enhanced plaque signal were identified as key risk factors for symptomatic plaques.

Conclusions:

• HR-VWI-based radiomics models effectively differentiate symptomatic from asymptomatic basilar atherosclerotic plaques.
• The developed radiomics approach demonstrates superior performance compared to traditional methods for identifying high-risk plaques.