P114 Radiomic features associated to clots composition in acute ischemic stroke

<h3>Introduction</h3> Approximately 30% of stroke are cryptogenic¹. The composition of clots can provide valuable insights into the possible etiology. No-invasive imaging such as radiomics can aid in assessing clot composition by studying very detailed imaging properties at a voxel level². <h3>Aim of Study</h3> Identify the radiomic profile of clot composition and its correlation with histology. <h3>Methods</h3> Patients with large vessel occlusion who underwent mechanical thrombectomy were included, and the extracted clots were analyzed using Micro-computed tomography (Micro-CT) imaging. Histological analysis was performed to obtain the percentage of red blood cells (RBCs) and fibrin (FB) within the clots. Histological slides were meticulously matched with corresponding areas on micro-CT images. Regions containing RBCs and FB were manually segmented using 3D Slicer. Radiomic features (RFs) were extracted for each region of interest. Student’s t-test and Pearson correlation were performed between RFs and clot composition. <h3>Results</h3> 130 RFs were extracted from ten clot specimens. Ten RFs (9 first-order and 1 second-order RFs) were different between RBCs and FB (p-value &lt; 0.05). One strong positive correlation and 7 strong negative correlations were found between RFs and the percentage of RBCs. Similarly, 12 strong positive correlations and 37 strong negative correlations were observed between RFs and FB composition. <h3>Conclusion</h3> Radiomics can potentially determine if a clot is primary composed of RBCs or FB.