Assessing the Accuracy of Software in Identifying Atherosclerotic Plaque Composition
Summary
Introduction
Carotid stenosis is a common cause for ischemic stroke and can be diagnosed through computed tomography angiography. Certain plaque characteristics are suggested to increase risk of ipsilateral ischemic stroke. This study aims to evaluate the accuracy of semi-automated software in identification of plaque tissue types by comparison with histology.
Methods
Patients, who had a preoperative computed tomography angiography and underwent carotid endarterectomy were included. Computed tomography angiographies were analyzed with semiautomated software to identify the composition of carotid plaques (dense calcium, fibrous tissue, fibrofatty tissue, and necrotic core). During pathological examination, plaque tissue types were scored (calcifications, collagen, fat, macrophages and intraplaque hemorrhage) and dichotomized. The software-derived and histological values were compared to evaluate the discriminative performance of the software.
Results
After exclusion of 59 patients, 44 were eligible for inclusion. Calcifications and macrophages could be significantly (p<0.05) discriminated on computed tomography angiography with an area under curve of 0.88 (95% CI 0.77-0.98) and 0.76 (95% CI 0.61-0.90) respectively. However, the area under curve of the discriminative ability of collagen and fat was 0.69 (95% CI 0.46-0.92) and 0.67 (95% CI 0.51-0.84) respectively, which did not achieve significance (p>0.05). The diagnostic accuracy for detecting the presence of intraplaque hemorrhage was low (area under curve = 0.40, 95% CI 0.19-0.60; p>0.05).
Conclusion
Semi-automated software has some, but limited accuracy in identification of plaque tissue types compared to histological evaluation. Further research should be conducted to more accurately determine the correspondence of semi-automated software with histology.