Using a CT Angiography Artificial Intelligence Carotid Plaque Morphology Model to Predict Stroke

The gold standard for determining carotid artery stenosis is to calculate using the North American Symptomatic Carotid Endarterectomy Trial criteria and ultimately plan medical vs surgical management based on percent symptomatic status. Few studies have assessed plaque morphology as an additive tool stroke prediction. Our study uses artificial intelligence software in conjunction with a patient's computed tomography (CT) scan of neck create three-dimensional (3D) model assess including calcification, intraplaque hemorrhage, matrix, perivascular adipose tissue. goal was predictive inclusive morphology. This retrospective review single tertiary institution from 2010 2021. Patients CT angiography head/neck diagnosis were included our analysis. Each run through third-party 3D image visibility We used stepwise backward regression select variables inclusion prediction models. Model discrimination receiver operating characteristic curves (AUC) slope. Additionally, calibration performed least Akaike information criterion (AIC) selected. primary outcome all cause mortality stroke. sample 366 patients over 11-year study. created three models predict mortality/stroke: A only clinical variables, B C both variables. age, sex, peripheral arterial disease, hyperlipidemia, body mass index, chronic obstructive pulmonary history transient ischemic attack or found AUC 0.737and AIC 285.4. tissue volume, cross sectional lumen area, calcified target lesion length 0.644 304.8. Finally, combined matrix attack/stroke, tissue, lipid rich necrotic core, hyperlipidemia 0.759 277.6 (Figs 1 2). demonstrate that combining factors creates best risk all-cause stenosis. Prospective are needed validate findings.Fig 2Model C: plot.View Large Image Figure ViewerDownload Hi-res Download (PPT)