Analysis of cerebral infarction pattern in computed tomography images of patients with internal carotid artery stenosis

An unbiased and quantitative analysis of lesion patterns in patient groups is described and applied to the analysis of infarction patterns. One hundred forty-two computed tomographic images of patients with ischemic stroke were registered to an average computed tomographic brain image, which was used as template. Lesions were segmented manually and averaged per category of internal carotid artery (ICA) stenosis. Thus, patterns of lesion distribution were formed. Differences then were analyzed using nonparametric statistics. This analysis was compared with the traditional classification of lesions. The nonparametric analysis showed an increased involvement of the territory of the middle cerebral artery in infarctions in patients with ICA occlusion compared with patients with mild or severe ICA stenosis. These differences did not show when classification of infarctions was used. The presented method is more sensitive to differences in lesion patterns than traditional lesion classification and showed a different infarction pattern for ICA occlusion