Directional Anisotropic Diffusion Applied to Segmentation of Vessels in 3D Images
Résumé
Anisotropic Diffusion is a new method derived from the convolution with a Gaussian, which allows to reduce the noise in the image without blurring the frontiers between different regions. This process can be applied in medical image analysis to segment the different anatomical structures. In this report, we introduce a new implementation of the anisotropic diffusion which allows us to reduce the noise and better preserve small structures like vessels in 3D images. This method is based on the differentiation of the diffusion in the direction of the gradient, and in the directions of the minimum and the maximum curvature. This algorithm gave good results on both synthetic and real images. We append to this work a part of the master's thesis of the first author (in French) which details several points of interest.