2026/07/01 by Junzhuo Li, Youxiang Cao, Feng Yang +5
Computer Science · Medicine · #Acute Ischemic Stroke Management #Cerebrovascular and Carotid Artery Diseases #Medical Image Segmentation Techniques
paper · doi:10.1002/ima.70410
openalex publication_date 2026/07/01 · openalex created_date 2026/07/15 · openalex updated_date 2026/07/29
ABSTRACT Accurate plaque localization is critical for stroke diagnosis, yet current clinical methods are subjective and experience‐dependent due to imaging limitations. This paper presents a multimodal MRI framework for automatic segmentation of cerebral arterial vasculature and atherosclerotic plaques, enabling early prevention. An automated framework for the segmentation of cerebral arteries and plaques based on multimodal MRI was constructed. Cerebral vascular structures were extracted using a Hessian‐based enhancement heatmap combined with a region‐selective segmentation method, ensuring the preservation of vascular details while eliminating enhanced regions. Subsequently, voxel registration between MRI and MRA to localize plaque search areas and a precise segmentation strategy integrating an improved SLIC algorithm with Dijkstra's shortest path algorithm was proposed. The proposed framework was validated on a dual‐center clinical dataset of 8 stroke patients, comprising 1152 MRA slices and 1664 MRI slices. The method achieves a Dice coefficient of 0.8966, an HD95 of 3.23 mm, an MCC of 0.8913, and an Accuracy of 0.9934. In comparison with 15 baseline methods, the proposed approach achieves superior performance across key metrics, including Dice, HD95, F1 Score, and MCC, with statistical significance ( p < 0.001 for most comparisons). The proposed multimodal MRI‐based framework for automated cerebral artery and plaque segmentation outperforms existing methods in accuracy and robustness. Through automated extraction of vascular and plaque, it alleviates the diagnostic workload for clinicians and provides a reliable tool for clinical diagnosis. The framework provides a reasonable and feasible technical means for cerebral arterial blood vessels and atherosclerotic plaque segmentation. It effectively addresses the issue of unintuitiveness in clinical diagnosis. It can provide support for the 3D quantitative analysis and evaluation of cerebral artery plaque.