Age-related diseases can be precipitated by metabolic disorder such as diabetes. In particular, vascular aging is characterized by profound modifications of large elastic arteries and vascular diseases involve remodeling of elastic lamellae in vessel walls. Our study aims to investigate arterial alterations due to aging and age-related diseases. Some characteristics of these alterations may help find biomarkers to forecast vascular outcome. In this study, we use synchrotron X-ray microtomography to acquire arterial wall images of mice of different ages and different health status. We segment and classify various regions of interest to analyze their modifications. We apply a methodological pipeline to extract the tunica media where elastic lamellae lie. We preprocess images to reduce noise. We further segment the lumen, which has a relatively clear boundary with the inner contour of the tunica media, by a co-occurrence texture analysis. Then, we compute a normal field to define image patches and orient them homogeneously to build a dataset used to train a Siamese neural network in order to find the outer contour of the tunica media. In this analysis, patches are labeled according to their similarities based on their spatial relations. We also propose a measure to evaluate the tortuosity of skeletonized elastic lamellae fragments to quantitatively compare their geometries in different health situations. The proposed pipeline leads to a segmentation precision higher than 90% on 2D cross-sections and the proposed measure shows a significant difference between the geometries of elastic lamellae in healthy and diabetic mice. Our current and future works are the segmentation of the lumen in a 3D view and a separation of the 5 different elastic lamellae with their enclosing structure straightened, in order to analyze their geometric features more accurately.