Background Intracranial atherosclerotic disease (ICAD) accounts for approximately 10% of acute ischemic strokes in the United States. While morphological characteristics such as the degree of stenosis have been used to identify high‐risk intracranial atherosclerotic plaques, the biological composition also plays a critical role in the pathogenesis of acute ischemic stroke. However, this type of analysis typically requires histopathological evaluation, which is only available post‐mortem. Radiomics, a voxel‐by‐voxel post‐processing imaging technique, enables quantification of signal intensity (SI). We aimed to characterize high‐risk plaque components, including the presence of macrophages and internal elastic lamina (IEL) disruption, using radiomics applied to high‐resolution magnetic resonance imaging (HR‐MRI). Methods Post‐mortem Circle of Willis (CoW) specimens were collected at our institution. Samples were fixed in 10% formalin and immersed in a 2% Polysorbate solution to remove blood clots. They were then embedded in agarose and imaged on a 3T GE MAGNUS scanner. Following image acquisition, specimens underwent histological analysis using hematoxylin and eosin staining, CD68 immunohistochemistry to identify macrophages, and elastichrome to detect IEL disruption. Imaging slices were matched with corresponding histological sections for each sample. Clusters of macrophages and IEL disruption were identified, and regions of interest were defined based on histological findings. Radiomic features (RFs) were then extracted, and univariate logistic regression was performed to identify RFs associated with these components. Results Five CoW specimens were included. Nine intracranial vessels were analyzed and 8/9 (88.8%) harbored atherosclerotic plaques while 1/9 (11.1%) had a healthy vessel wall. Histological analysis showed that 7/8 (88%) plaques had macrophages and 6/8 (75%) plaques had IEL disruption. The RFs associated to macrophages included Difference Entropy (OR: 4.22, 95% CI: 1.40‐12.74, p = 0.01), Joint Entropy (OR: 3.10, 95% CI: 1.20‐7.98, p = 0.01) and Sum Entropy (OR: 3.63, 95% CI: 1.29‐10.30, p = 0.01). The RFs correlated with IEL disruption were Inverse Variance (OR: 1.05, 95% CI: 1.01‐1.10, p = 0.03), Joint Entropy (OR: 1.07, 95% CI: 1.02‐1.13, p = 0.006), Maximum Probability (OR: 1.07, 95% CI: 1.01‐1.13, p = 0.008) and Small Area Low Gray Level Emphasis (OR: 1.05, 95% CI: 1.01‐1.11, p = 0.04). Conclusions Radiomics is a promising non‐invasive imaging tool for characterizing components of intracranial atherosclerotic plaques in HR‐MRI. Figure: Ex Vivo Imaging of Atherosclerotic Plaques. (A) A basilar artery showing an IEL disruption on histology with corresponding HR‐MRI slide. (B) A posterior cerebral artery depicting macrophages on histology and the consequent slide on HR‐MRI. image
Gudino et al. (Sat,) studied this question.