Abstract 12: CT Perfusion For Lesion-Symptom Mapping In Emergent Large Vessel Occlusion

Introduction: Selection criteria for endovascular thrombectomy (EVT) for emergent large-vessel occlusion (ELVO) stroke increasingly rely on CT perfusion (CTP) findings at presentation. While perfusion imaging is used to estimate the likelihood of deriving long-term functional benefit from EVT, there is comparatively little data on how perfusion deficits relate to domains of neurological function. Here, we investigated region-specific hypoperfusion as a driver of neurological dysfunction through correspondence between NIHSS items and CTP. Methods: We included 169 patients with ELVO presenting to Mayo Clinic Rochester or Jacksonville, who had baseline CTP and itemised NIHSS data. Perfusion was quantified as mean transit time using e-CTP (Brainomix Ltd., Oxford, UK) and thresholded to ≥5 s for lesion masking. Voxel-wise lesion-symptom mapping (VLSM) was performed using sparse canonical correlations analysis (Lesymap package, R). Results: Total NIHSS was most strongly predicted by left MCA territory hypoperfusion, corresponding to language-eloquent cortical areas and dominant hemisphere motor tracts (figure 1). Contralateral limb motor deficits (NIHSS 5 and 6) were mainly associated with subcortical and white matter hypoperfusion; there was an additional strong signal for left frontal cortex in language function (NIHSS 9). The remaining NIHSS items were less well localised. Conclusions: Expanding on previous pilot data, our results provide further evidence for the feasibility of hypoperfusion-to-symptom mapping in ELVO. This could allow for more nuanced clinical-imaging mismatch determination, and selection of patients based on specific neurological deficits that may be reversible.