Skip to Main content Skip to Navigation

Segmentation d'IRM cérébrales multidimensionnelles par coupe de graphe

Jérémy Lecoeur 1
1 VisAGeS - Vision, Action et Gestion d'informations en Santé
INSERM - Institut National de la Santé et de la Recherche Médicale : U746, Inria Rennes – Bretagne Atlantique , IRISA-D5 - SIGNAUX ET IMAGES NUMÉRIQUES, ROBOTIQUE
Abstract : This thesis deals with the segmentation of multimodal brain MRIs by graph cuts method. First, we propose a method that utilizes three MRI modalities by merging them. The border information given by the spectral gradient is then challenged by a region information, given by the seeds selected by the user, using a graph cut algorithm. Then, we propose three enhancements of this method. The first consists in finding an optimal spectral space because the spectral gradient is based on natural images and then inadequate for multimodal medical images. This results in a learning based segmentation method. We then explore the automation of the graph cut method. Here, the various pieces of information usually given by the user are inferred from a robust expectation-maximization algorithm. We show the performance of these two enhanced versions on multiple sclerosis lesions. Finally, we integrate atlases for the automatic segmentation of deep brain structures. These three new techniques show the adaptability of our method to various problems. Our different segmentation methods are better than most of nowadays techniques, speaking of computation time or segmentation accuracy.
Complete list of metadatas

Cited literature [10 references]  Display  Hide  Download
Contributor : Jérémy Lecoeur <>
Submitted on : Thursday, July 15, 2010 - 7:04:28 PM
Last modification on : Thursday, January 14, 2021 - 11:16:47 AM
Long-term archiving on: : Tuesday, October 23, 2012 - 10:25:57 AM


  • HAL Id : tel-00502842, version 1


Jérémy Lecoeur. Segmentation d'IRM cérébrales multidimensionnelles par coupe de graphe. Informatique [cs]. Université Rennes 1, 2010. Français. ⟨tel-00502842⟩



Record views


Files downloads