Investigation of cerebral vascular abnormalities in TOF-MRA brain images

Abstract

Abnormalities in the arterial system can lead to a series of cerebrovascular diseases including unruptured intracranial aneurysms (UIAs). UIAs are present in around 3% of the population, and upon rupture, 35% of cases result in death and most of the survivors are left with long-term disabilities. Aneurysms are usually small in size and vary greatly in shape and position in the vessel configuration. It is one of the reasons why up to 10% of UIAs are missed during screening. For this reason, in this study, an automatic atlas-based method of at-risk-areas and aneurysm detection is presented. The pipeline outputs a color-coded map that can be superimposed on the TOF-MRA image, indicating areas that diverge from the average healthy anatomy in the form of Z-scores. High Z-scores are potentially linked to the presence of abnormalities. This method was tested on 19 TOF-MRA containing one aneurysm above 5 mm in radius and 18 images with healthy anatomy. After preprocessing, the TOF-MRA underwent vessel segmentation and vessel radius calculation. The resulting image, as well as the preprocessed atlases registered to the patient’s space, are then used to calculate the Z-score maps. Using two different types of segmentation for evaluation, aneurysms were detected in up to 17 images out of 19. An analysis of the Z-scores in areas outside the segmentations showed no statistical differences compared to the Z-score map from healthy subjects. This method has the potential to be useful in a clinical setting, as well as research, as part of larger projects.