Assessing Middle Cerebral Artery Atherosclerotic Plaque Vulnerability Using a 3D HRMR-VWI-Based Radiomics and Fractal Dimension Fusion Model

Abstract Objective To investigate the value of Fractal Dimension (FD) as a novel quantitative imaging biomarker in discriminating vulnerable atherosclerotic plaques of the middle cerebral artery (MCA) and to evaluate the diagnostic value of its integration with radiomic features. Methods This retrospective study included 194 patients with MCA atherosclerosis who underwent HRMR-VWI at Shanxi Cardiovascular Hospital from September 2021 to December 2023. Based on clinical symptoms and DWI findings, they were categorized into a symptomatic plaque group (n = 111) and an asymptomatic plaque group (n = 83). Radiomic features and fractal dimension parameters (FD-Grayscale, FD-Binary, Lacunarity) were extracted from non-contrast and contrast-enhanced 3D T1WI images. Feature selection was performed using Minimum Redundancy Maximum Relevance (MRMR) and LASSO regression, with logistic regression as the classifier. Predictive models were constructed, including FD-only models, radiomics-only models, and an integrated FD-Rad model using features from non-contrast, contrast-enhanced, or combined images. Model performance was assessed using the Area Under the Receiver Operating Characteristic Curve (AUC), calibration curves, and Decision Curve Analysis (DCA). Results FD-Grayscale, FD-Binary, and Lacunarity were significantly higher in the symptomatic group than in the asymptomatic group (all P < 0.001). The integrated FD-Rad model demonstrated the best performance in the test set, with an AUC of 0.917, accuracy of 0.831, and specificity of 0.880, along with good calibration and clinical net benefit. The radiomics model generally exhibited high sensitivity, while the FD model showed high specificity, indicating complementary diagnostic characteristics. Correlation analysis revealed that FD-Binary was significantly correlated with some texture features, whereas FD-Grayscale showed weak correlation, confirming that fractal dimension provides unique diagnostic information. Conclusion The fusion model integrating 3D HRMR-VWI-based radiomics and fractal dimension significantly improves the efficacy in discriminating vulnerable MCA plaques. Fractal dimension features, serving as novel quantitative metrics for characterizing plaque morphological complexity and spatial heterogeneity, possess interpretability and clinical translation potential. This approach provides a novel perspective and complementary information for vulnerability assessment, holding promise as an effective tool for precisely identifying high-risk patients and guiding clinical intervention strategies.