Three-Dimensional Reconstruction of Carotid Arteries Using Computed Tomography Angiography

Abstract

Carotid artery disease is considered as the pathological disease of carotid arteries and is considered as a principal cause of stroke. Therefore, early diagnosis of carotid artery disease is of high clinical importance. This study aims to present an overall methodology for the accurate identification of the inner wall, outer wall and the atherosclerotic plaques (calcified and non-calcified) of the carotid arteries. The proposed methodology is based on a level set based approach, which is fully adapted in each computed tomography acquisition protocol. Briefly, the methodology includes the following steps: (i) the estimation of intensity membership functions for the inner wall, the outer wall and CP, (ii) the carotid artery centerline extraction, (iii) the inner wall, outer wall and calcified plaques segmentation, (iv) the noncalcified plaques segmentation and finally (v) the 3D models construction. The segmentation accuracy of the proposed methodology has been validated against manual expert's annotations in 4 patients, and more specifically in 300 computed tomography angiography slices for the inner wall segmentation and in 30 slices as far as the atherosclerotic plaques is concerned. The utilized evaluation metrics were the Dice coefficient and the Hausdorff Distance and our very first results are promising for the accurate and automated segmentation of carotid arteries.