Skip to Main content Skip to Navigation

Reconnaissance automatique des sillons corticaux

Abstract : The determination of specic biomarkers of brain pathologies at population scale is extremely dicult because of the huge inter-individual variability of the sulco-gyral topography. This thesis addresses this issue by automatically identifying 125 sulcal structures and pairing them through individuals, thanks to a manually labeled database of 62 subjects. Relying on the sulcal roots theory, cortical folds are split into elementary segments to be labeled. In a rst time, the structural approach proposed earlier by Jean-François Mangin and Denis Rivière has been revisited to manage the increasing amount of morphometric features involved in the identication process. In a second time, this model has been fully reviewed in favor of a Bayesian framework based on localized information (positions or directions) previously neglected, thus allowing eective optimization schemes. In this context, data normalization is essential ; this issue has been considered through global or sulciwise local ane registration techniques, jointly to the sulcal identication. In order to introduce more structural informations, a Markovian model has been successfully introduced to reect the local neighbored cortical folds organization. Finally, the overall recognition rate has reached 86 % for each hemisphere. From now on, only atypical patterns or the most variable anatomical structures remain a real issue.
Document type :
Complete list of metadatas

Cited literature [18 references]  Display  Hide  Download
Contributor : Ens Cachan Bibliothèque <>
Submitted on : Tuesday, February 16, 2010 - 3:32:16 PM
Last modification on : Thursday, July 2, 2020 - 5:17:18 PM
Long-term archiving on: : Thursday, October 18, 2012 - 3:11:43 PM


  • HAL Id : tel-00457072, version 1



Matthieu Perrot. Reconnaissance automatique des sillons corticaux. Mathématiques [math]. École normale supérieure de Cachan - ENS Cachan, 2009. Français. ⟨tel-00457072⟩



Record views


Files downloads