Evaluation of Computer-Assisted Quantification of Carotid Stenosis

Accurate segmentation of the internal carotid artery (ICA) is a critical step towards the diagnosis of carotid artery disease. A method is presented and evaluated for segmentation of internal carotid arteries using adaptive thresholding. Thirty-two arteries were segmented with the proposed method and their stenoses classified. A specialised graphical user interface tool for 3D artery visualization was created to enable manual delineation and segmentation. The segmentation results from two clinical experts using this tool were used for evaluation of the automatic method. The comparison of the manual and automated delineations show that there is a good correspondence. Treating the automatic method as a third "observer", inter-observer variability was also evaluated, and it was found that the algorithm's performance was comparable to that of a clinical expert.