Modeling the 3D coronary tree for labeling purposes

Angiographic Image Analysis

“…In the context of angiographic imaging, we consider that vessel segmentation embeds (i) methods that detect either whole vessels (i.e., their lumen and/or walls) or their me dial axes, and/or (ii) methods that perform lowlevel processing or highlevel knowl edge extraction (e.g., vein/artery discrimination [100,103] or vessel labelling [14,42]). We also consider some methods which could be classified as filtering ones, since their purpose is to perform vessel enhancement, which consists mainly of denoising, but also of vessel reconnection (for example in case of stenosis, or of signal loss [27,76]).…”

“…Such a highly semantic task can not, of course, be carried out without using highlevel a priori anatomical knowledge. Based on these considerations, several vascular atlases have been involved in -and sometimes specifically designed for-this labelling task [14,30,36,41,42].…”

“…This is the approach proposed in [14] for generating a graphbased atlas of the coronary arteries. The use of a phantom enables to easily obtain a segmentation (which can be validated a posteriori) and to derive, by a topological postprocessing, curvilinear structures en abling to define a graph structure.…”

“…The use of a phantom enables to easily obtain a segmentation (which can be validated a posteriori) and to derive, by a topological postprocessing, curvilinear structures en abling to define a graph structure. In [14], such an atlas can be achieved by storing at each edge/vessel segment information attributes such as its name, length, orientation and diameter. Geometrybased atlases The works described above have focused on vascular struc tures presenting simple properties, namely the coronary arteries, out of their anatomical neighbouring context.…”

Assessment of qualitative and quantitative features in coronary artery MRA

“…In both cases, imaging sources play an important role as it enables to detect and characterize potential pathological patterns and to elaborate sound planning for therapy and efficient patient followup [5]. They provide relevant anatomical knowledge of the human heart vasculatures (either morphometric [6,7], structural [8,9] or both [10]) which can be confronted to patient-specific data in a therapeutic or surgical decisional process. Despite the high resolution of the X-Ray angiography, the gold standard for coronary analysis, and the many attempts in reconstruction from few views [11,12,13], other techniques are increasingly being used in clinical protocols : Multidetector Computed To-…”