INTRODUCTION: Endovascular thrombectomy (EVT) is the gold standard treatment for managing emergent large vessel occlusion (LVO) ischaemic stroke. The role of perfusion CT imaging in selecting appropriate candidates for EVT is well established. New understanding of the benefits of EVT in later stages of infarct have raised questions about the requirement of perfusion CT. This study assessed the accuracy of non-perfusion imaging for making appropriate EVT referrals. METHODS: We reviewed 50 consecutive acute stroke presentations with-contrast CT, CTA and perfusion imaging. Observers included two neuroradiologists and two senior radiology registrars. Cases were anonymised, and perfusion data removed. Diagnostic accuracy metrics (sensitivity, specificity, PPV, NPV) for evaluating hyperdense vessel (HV), acute infarct (AI) and LVO were calculated. Interobserver agreement was measured. Sensitivity for recommending IR discussion was compared against clinical outcomes. RESULTS: High interobserver agreement was observed for LVO detection (κ = 0.92) and IR referral (κ = 0.96). Consultants demonstrated higher sensitivity for HV (92%) and AI (76%), while registrars exhibited higher specificity (HV: 94%; AI: 96%). Both groups achieved > 90% accuracy in LVO detection. Sensitivity for IR recommendations was 91.5% for consultants and 96.0% for registrars (p = 0.5). All cases that proceeded for EVT in this cohort were identified by all four reviewers and the same clinical recommendation was made. CONCLUSION: Non-perfusion imaging offers robust diagnostic accuracy for detecting LVO, making it a reliable tool for EVT decision-making in resource-constrained centres. Perfusion imaging enhances diagnostic confidence and clinical planning, particularly in complex cases. Its integration into workflows should remain a priority where feasible.
Sunderland et al. (Wed,) studied this question.