Abstract
Advances in imaging technologies and computational analyses have enabled the comprehensive evaluation of cerebrovascular morphology, thereby facilitating the identification of predictive indicators of cerebrovascular events. This study evaluated the association between the cerebral arterial morphological features and the incidence of ischaemic stroke while considering intracranial atherosclerotic stenosis and vascular distribution. This retrospective study analysed data from patients with acute ischaemic stroke or transient ischaemic attack between January 2017 and December 2020 who underwent time-of-flight magnetic resonance angiography at least twice over a minimum interval of 3 years. Data were analysed between September 2003 and November 2023. Arterial features, including baseline measurements and annual changes, were quantified across the entire cerebrovascular system using in-house vessel analysis software. Kaplan-Meier survival analysis and Cox proportional hazards regression were performed to assess the association between the arterial features and ischaemic stroke stratified according to the presence of intracranial atherosclerotic stenosis. Of the total study population of 462 patients (mean age 64.35 ± 13.22 years, 64.29% male; mean follow-up 5.28 ± 2.58 years, up to 10 years), 235 experienced acute ischaemic stroke during the follow-up period. Several global features associated with ischaemic stroke were identified. The baseline minimum-maximum diameter ratio and luminal circularity exhibited positive correlations with ischaemic stroke, whereas the baseline curvature exhibited a negative correlation. A chunk-based feature analysis, which accounted for the vascular distribution, was conducted to explore these associations. The baseline minimum-maximum diameter ratio exhibited positive correlations primarily in small arteries. In contrast, the baseline curvature exhibited negative correlations in specific regions, such as parts of the posterior circulation and the left middle cerebral artery. The subgroup analysis revealed differing patterns of association based on the presence of intracranial atherosclerotic stenosis. The correlation patterns observed in the total population were largely preserved in the group without stenosis, whereas no significant correlations were identified in the group with stenosis. These findings suggest that intracranial atherosclerotic stenosis may obscure these relationships. Thus, the present study highlights the importance of integrating whole-brain vascular morphology analysis with the stratification of intracranial atherosclerotic stenosis to understand the mechanisms of ischaemic stroke. These findings provide insights into potential predictive metrics that will contribute to the development of advanced models for personalized stroke prevention and management.