Model-Based Segmentation and Recognition of Anatomical Brain Structures in 3D MR Images
We propose a coarse to fine framework for the segmentation and recognition of structures in brain MR Images. We first coarsely segment the outer surface of the brain, which is used as the main criteria for the non-rigid registration of the MR image with a Computerised Brain Atlas. Then, we use the structures in the atlas as the initialisation for active surfaces models. Those surfaces are deformed in order to minimise an energy depending both on the smoothness of the structure and on the image itself.
KeywordsActive Contour Model Global Transformation Smoothness Term Euclidean Distance Transform Computerise Brain
Unable to display preview. Download preview PDF.
- O. Cuisenaire, J.P. Thiran, B. Macq, C. Michel, A. De Voider and F. Marques. Automatic registration of 3D MR images with a computerised brain atlas, SPIE Medical Imaging 96, Newport Beach, SPIE vol. 1719, pp. 438–449Google Scholar
- O. Cuisenaire. Region Growing Euclidean Distance Transforms. ICIAP’97, Florence, September 97, Lecture Notes in Computer Science.Google Scholar
- L D Cohen and I. Cohen. Finite element methods for active contour models and balloons for 2D and 3D images. IEEE Pattern Analysis and Machine Intelligence, 15, November 1993Google Scholar
- W.E. Lorensen and H.E. Cline. Marching Cubes: a High Resolution 3D Surface Construction Algorithm. Proceedings of SIGGRAPH’87, Computer Graphics, vol. 21, No. 4, pp. 163–169.Google Scholar