Skip to Main content Skip to Navigation

Contributions à la segmentation des structures cérébrales en IRM foetale

Abstract : The study of cerebral maturation aims at a better understanding of the brain's development during the pregnancy and the high lightment of the links between the change in cerebral structures and the cognitive development. This study is particularly difficult because of the constant evolution of these structures during the pregnancy. This evolution is due to the growing and the organisation of the different cerebral tissues. The preferred visualisation technique used to observe the brain is the magnetic resonance imaging (MRI), which is a non invasive acquisition technique of in vivo 3D cerebral images with a relative high resolution. Nevertheless, significant anatomical differences and the fast evolution of the fetal cerebral structures require a new modelisation of the brain. This thesis is divided in two parts. First, we modified the Fuzzy C-Means (FCM) algorithm in order to allow a better consideration of images noise and intensity bias thanks to the non-local means method, originaly elaborated for denoising purposes. These work were validated with databases of simulated and real images. Finally, we focused on the segmentation of cerebral tissues in fetal MRI by introducing topological constraints in order to obtain a sequential segmentation, based on relative positions of the tissus. These work were validated with real cases.
Complete list of metadatas
Contributor : Abes Star :  Contact
Submitted on : Friday, November 2, 2012 - 1:02:15 PM
Last modification on : Wednesday, August 22, 2018 - 1:10:06 PM
Long-term archiving on: : Sunday, February 3, 2013 - 3:36:47 AM


Version validated by the jury (STAR)


  • HAL Id : tel-00747860, version 1



Benoît Caldairou. Contributions à la segmentation des structures cérébrales en IRM foetale. Traitement du signal et de l'image [eess.SP]. Université de Strasbourg, 2012. Français. ⟨NNT : 2012STRAD015⟩. ⟨tel-00747860⟩



Record views


Files downloads