Skip to Main content Skip to Navigation

Fusion d'images 3D du cerveau - Etude de modèles et applications

Abstract : The collection of various data coming from imagery, expert knowledge or physiological signals is becoming very common for the study of a given pathology. The treatment of these data is performed by a physicist, who analyses and aggregates them according to his knowledge, and generally leads to a medical decision. The aim of this work is to model this aggregation process by means of a fusion technique, in the case of brain studies. The fusion process is divided into three steps: We first model the available information, numerical or symbolic, in a common theoretical frame. The possibilistic logic allows for the management of ambiguities and imprecision that are inherent to medical data. We thus propose to model on the one hand the distribution of cerebral tissues in anatomic (MR) and functional (SPECT and TEP) images by means of a fuzzy clustering algorithm on appropriate feature vectors, and on the other hand the symbolic information coming from expert knowledge. We then aggregate these information with a fusion operator. This operator has to affirm redundancy, manage the complementarities and also take into account conflicts, that often underline the presence of a pathology. We thus propose three models for three precise clinical cases: the fusion of MR images, the fusion of anatomical and functional images and the fusion of MR images with symbolic information. We finally propose a synthetic piece of information that allows to best represent the available data. We define for the threee previous models a resulting image that allows for example to propose a diagnosis, to establich a prognosis or to provide help for a surgical planning. Four clinical applications illustrate these concepts: brain tissue volumes quantification, study of Alzheimer's disease, study of epilepsy and segmentation of the subthalamic nucleus in the treatment of Parkinson's disease. For every case, besides the basic model previously described, we propose specific treatments and a clinical validation. An industrial application in collaboration with SEGAMI corporation, that finalizes and industrially increases this work, is finally presented.
Complete list of metadatas
Contributor : Vincent Barra <>
Submitted on : Wednesday, April 28, 2004 - 9:01:08 AM
Last modification on : Monday, January 21, 2019 - 10:26:03 AM
Long-term archiving on: : Friday, April 2, 2010 - 9:09:54 PM


  • HAL Id : tel-00005976, version 1



Vincent Barra. Fusion d'images 3D du cerveau - Etude de modèles et applications. Autre [q-bio.OT]. Université d'Auvergne - Clermont-Ferrand I, 2000. Français. ⟨tel-00005976⟩



Record views


Files downloads