ABSTRACT NUMBER: ESOC2026A692 RADIOLOGICAL DETERMINANTS OF TISSUE FATE IN PLAIN BRAIN COMPUTER TOMOGRAPHY DURING ACUTE LARGE VESSEL OCCLUSION: A MULTICENTER STUDY Authors: Haipeng Li, Billy Lam, Bianca Chan, Sze Ho MA, Trista Hung, Ho Ko, Hao Wang, Junjie Sai, Thomas Leung, Bonaventure Ym Ip Published: 2026 ABSTRACT AbstractBackground and aimsAlberta Stroke Program Early CT Score (ASPECTS), a semiquantitative assessment, is increasingly adopted on endovascular thrombectomy (EVT) triage. However, the relative contribution of components of baseline ASPECTS (hypoattenuation, loss of grey-white differentiation [LGWD], and sulcal effacement or edema in subcortical structures [SE]) to tissue fate in patients with large vessel occlusion (LVO) undergoing EVT is uncertain. We aimed to elucidate the association between specific ASPECTS features and final infarct volume (FIV).MethodsWe conducted a multicenter retrospective cohort study across 4 hospitals in Hong Kong and mainland China. Pre-EVT ASPECTS was assessed by experienced neuroradiologists and automated software on plain CT brain. Hypoattenuation, LGWD and SE were graded manually for each ASPECTS region. The primary outcome was the FIV on post-EVT MRI. The relationship between each ASPECTS component and FIV was determined by using multivariable regression models. A weighted ASPECTS model was then constructed based on the effect estimate for each feature.ResultsAmong 350 patients with LVO undergoing successful EVT between January 2020 and July 2024, hypoattenuation has the most robust relationship, which is significantly associated with FIV across all ASPECTS regions, while sulcal effacement has the least (Figure 1). The weighted ASPECTS (β-coefficient = -0.77 [-0.89, -0.65], p<0.001) demonstrated a stronger association with FIV than neuroradiologist-graded (β-coefficient = -0.54 [-0.63,-0.45], p<0.001) or AI-graded ASPECTS (β-coefficient = -0.58 [-0.70, -0.46], p<0.001) (Figure 2).ConclusionsASPECTS regions with different early ischemic features differ in their predictive strength for FIV. Weighted feature-informed ASPECTS may improve the final infarct prediction and EVT patient selection.Conflict of interestNothing to discloseFigure 1 - belongs to Results Figure 2 - belongs to Results