Skip to Main content Skip to Navigation

Trois principes de coopération pour la segmentation en imagerie de résonnance magnétique cérébrale

Laurence Germond 1
TIMC - Techniques de l'Ingénierie Médicale et de la Complexité - Informatique, Mathématiques et Applications, Grenoble - UMR 5525
Abstract : The goal of our work is the conception and implementation of a segmentation system for Magnetic Resonance Images of the brain. The aim is to combine complementary approaches in response to the complexity of the problem. Given the diversity of the approaches existing for segmentation (classification, deformable models,...) we demonstrate that it is possible to define a common framework to allow cooperation between heterogeneous approaches. The methodological framework we have defined is based on learning and adaptability. Learning allows for the introduction of a priori explicit knowledge about the context under the form of a statistical deformable model of the brain boundary. The resulting information is used to specialise locally agents of segmentation, and thus provides adaptability. We have designed three principles of cooperation to perform the combination of information : (i) cooperation through initialisation, that exploits the result of one method to initialise another method, (ii) cooperation through fusion of information, that allows to take advantage of the fusion of complementary sources of information to guide a stage of processing, (iii)cooperation through retroaction, that allows to retroact on the whole process of segmentation either to refine it or to transmit information between images. The whole work has been validated on synthetic data provided by the Montreal Neurological Institute and also illustrated on a set of real images.
Document type :
Complete list of metadata

Cited literature [77 references]  Display  Hide  Download
Contributor : Thèses Imag <>
Submitted on : Wednesday, February 18, 2004 - 2:57:04 PM
Last modification on : Thursday, November 19, 2020 - 3:56:09 PM
Long-term archiving on: : Friday, April 2, 2010 - 8:27:25 PM


  • HAL Id : tel-00004835, version 1



Laurence Germond. Trois principes de coopération pour la segmentation en imagerie de résonnance magnétique cérébrale. Interface homme-machine [cs.HC]. Université Joseph-Fourier - Grenoble I, 1999. Français. ⟨tel-00004835⟩



Record views


Files downloads