Aortic and carotid vessel wall MRI have been applied in clinical trials to monitor the progression and regression of atherosclerosis by assessing the wall thickness (WT). The analysis of these data requires vessel wall segmentation and several studies have reported automated methods^[1,2,3]. However, most of the methods require user interaction as an input to the automated process. In addition, all these segmentation algorithms are dedicated to either the carotid artery or the aorta. Aim: To develop a method that can fully automatically segment the vessel wall and calculate WT for both the common carotid artery and the descending aorta in MR images, which is especially suitable for cohort studies where the detection of temporal WT changes is needed in a large dataset.MRI: In 10 healthy volunteers repeated MRI scans of the carotid artery were acquired and in another 10 healthy volunteers repeated MRI scans of aorta. In 10 patients with essential hypertension and 10 age-matched and sex-matched healthy volunteers one MRI scan of both the carotid artery and the aorta was acquired. All vessel wall images were acquired using a 3.0T scanner with an ECG triggered DIR (black-blood) segmented k-space gradient-echo sequence with fat suppression. Cross-sectional images of the left common carotid were acquired using the following parameters: TR/TE = 12/3.6 ms, FA= 45◦, FOV = 140×140 mm², resolution = 0.27×0.27×2 mm³. Cross-sectional images of descending aorta were acquired with: TR/TE = 4.9/2.5 ms, FA = 20◦, FOV = 270×270 mm², resolution = 0.53×0.53×2 mm³. Automated segmentation: The proposed algorithm includes the following four steps which are all performed automatically: Step 1: localization of the vessel of interest based on the circular Hough Transform (CHT) (Figure 1 b, e); Step 2: estimation of the lumen center and radius using oriented-edge based Hough Transform (OEHT) (Figure 1 c, f); Step 3: initialization of the tube model for lumen followed by detection and refinement of the lumen boundary based on a tube-fitting method^[1] and dynamic programing^[2]; Step 4: initialization of the tube model for outer vessel wall based on the dilation of lumen segmentation followed by segmentation of the outer wall using a similar manner as for lumen. Manual segmentation: Two trained radiologists (observer 1 and observer 2) delineated the vessel wall contours using VesselMASS software^[2]. The delineations performed by observer 1 served as the gold standard, and those made by observer 2 were used to quantify inter-observer variability. Evaluation: WT was calculated by taking the median length of 100 chords connecting lumen and outer wall boundaries^[1]. 1) To directly compare the manual and automatic generated contours, the mean contour distance (MCD) was calculated by taking the average of the distances between the two sets of contour points. 2) To evaluate the performance and inter-scan reproducibility of the automated WT assessment, Bland-Altman analysis and paired t-test were performed. 3) To compare the WT between the healthy volunteers and hypertensive patients, unpaired T-test was computed. For all statistical tests, a P < 0.05 was considered to be statistically significant.Generally, all the mean differences between the manual and automated segmentation results were smaller than one pixel (Table 1). The vessel wall thickness was slightly, but significantly underestimated by the automated segmentation for carotid and aorta; however, the difference between observer 1 and observer 2 was also statistically significant (Table 1). For both carotid and aorta, no significant difference was observed between the inter-scan wall thickness assessments using automated segmentation (Table 1). Automated segmentation detected significantly higher carotid and aortic wall thickness in the hypertensive patients (Table 2). We presented a fully automated contour detection and vessel wall thickness quantification method. The results demonstrated that our method has good agreement with manual segmentation, excellent scan-rescan reproducibility and capability of detecting vessel wall thickening in hypertensive patients. Therefore, the presented method is promising for application in future cohort studies.