Segmentation of tumor vessels based on parallel double snakes including region information

Pierre Glanc (1) , Florence Rossant (1) , Valérie Rouffiac (2) , Corinne Laplace-Builhé (2) , Ingrid Leguerney (2) , Stéphanie Pitre-Champagnat (2)

1 ISEP - Institut Supérieur d'Electronique de Paris
2 IR4M - Imagerie par Résonance Magnétique Médicale et Multi-Modalités

In this paper, we address the problem of the seg-mentation of vessels in images of mouse tumors, with an efficient algorithm that minimizes the user's intervention. For each vessel, two points delimiting its extremities have to be selected. Then, a line inside the vessel is automatically determined based on a Dijkstra-type algorithm. Finally, an original active contour model combining both parallel double snakes and region criteria aims at finding the borders of the vessel. Our segmentation algorithm provides numerical models of tumor vessels, suitable for the simulation of blood and contrast agent flow.

Mots clés: Image segmentation; Tumor vessels; Active contour models; Parallel Double Snakes; Region criteria

Domaines: Traitement du signal et de l'image [eess.SP]; Imagerie médicale; Imagerie

Pierre Glanc, Florence Rossant, Valérie Rouffiac, Corinne Laplace-Builhé, Ingrid Leguerney, et al.. Segmentation of tumor vessels based on parallel double snakes including region information. 5th International Conference on Image Processing Theory, Tools and Applications (IPTA'15), Nov 2015, Orléans, France. pp.174-179, ⟨10.1109/IPTA.2015.7367121⟩. ⟨hal-01568726⟩