Afficher la notice abrégée

AssemblyNet: A large ensemble of CNNs for 3D whole brain MRI segmentation

hal.structure.identifierLaboratoire Bordelais de Recherche en Informatique [LaBRI]
hal.structure.identifierPatch-based processing for medical and natural images [PICTURA]
dc.contributor.authorCOUPÉ, Pierrick
hal.structure.identifierLaboratoire Bordelais de Recherche en Informatique [LaBRI]
dc.contributor.authorMANSENCAL, Boris
hal.structure.identifierLaboratoire Bordelais de Recherche en Informatique [LaBRI]
hal.structure.identifierPatch-based processing for medical and natural images [PICTURA]
dc.contributor.authorCLÉMENT, Michaël
hal.structure.identifierLaboratoire de l'intégration, du matériau au système [IMS]
dc.contributor.authorGIRAUD, Rémi
hal.structure.identifierInstitut de Mathématiques de Bordeaux [IMB]
hal.structure.identifierModélisation Mathématique pour l'Oncologie [MONC]
dc.contributor.authorDENIS DE SENNEVILLE, Baudouin
hal.structure.identifierInstitut Polytechnique de Bordeaux [Bordeaux INP]
hal.structure.identifierPatch-based processing for medical and natural images [PICTURA]
dc.contributor.authorTA, Vinh-Thong
hal.structure.identifierLaboratoire Bordelais de Recherche en Informatique [LaBRI]
dc.contributor.authorLEPETIT, Vincent
hal.structure.identifierITACA
dc.contributor.authorMANJÓN, José
dc.date.accessioned2024-04-04T02:50:07Z
dc.date.available2024-04-04T02:50:07Z
dc.date.issued2020-10
dc.identifier.issn1053-8119
dc.identifier.urihttps://oskar-bordeaux.fr/handle/20.500.12278/191883
dc.description.abstractEnWhole brain segmentation of fine-grained structures using deep learning (DL) is a very challenging task since the number of anatomical labels is very high compared to the number of available training images. To address this problem, previous DL methods proposed to use a single convolution neural network (CNN) or few independent CNNs. In this paper, we present a novel ensemble method based on a large number of CNNs processing different overlapping brain areas. Inspired by parliamentary decision-making systems, we propose a framework called AssemblyNet, made of two "assemblies" of U-Nets. Such a parliamentary system is capable of dealing with complex decisions, unseen problem and reaching a relevant consensus. AssemblyNet introduces sharing of knowledge among neighboring U-Nets, an "amendment" procedure made by the second assembly at higher-resolution to refine the decision taken by the first one, and a final decision obtained by majority voting. During our validation, AssemblyNet showed competitive performance compared to state-of-the-art methods such as U-Net, Joint label fusion and SLANT. Moreover, we investigated the scan-rescan consistency and the robustness to disease effects of our method. These experiences demonstrated the reliability of AssemblyNet. Finally, we showed the interest of using semi-supervised learning to improve the performance of our method.
dc.description.sponsorshipApprentissage profond pour la volumétrie cérébrale : vers le BigData en neuroscience
dc.description.sponsorshipTranslational Research and Advanced Imaging Laboratory - ANR-10-LABX-0057
dc.language.isoen
dc.publisherElsevier
dc.title.enAssemblyNet: A large ensemble of CNNs for 3D whole brain MRI segmentation
dc.typeArticle de revue
dc.identifier.doi10.1016/j.neuroimage.2020.117026
dc.subject.halInformatique [cs]/Imagerie médicale
bordeaux.journalNeuroImage
bordeaux.page117026
bordeaux.volume219
bordeaux.hal.laboratoriesInstitut de Mathématiques de Bordeaux (IMB) - UMR 5251*
bordeaux.institutionUniversité de Bordeaux
bordeaux.institutionBordeaux INP
bordeaux.institutionCNRS
bordeaux.peerReviewedoui
hal.identifierhal-02930959
hal.version1
hal.popularnon
hal.audienceInternationale
hal.origin.linkhttps://hal.archives-ouvertes.fr//hal-02930959v1
bordeaux.COinSctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.jtitle=NeuroImage&rft.date=2020-10&rft.volume=219&rft.spage=117026&rft.epage=117026&rft.eissn=1053-8119&rft.issn=1053-8119&rft.au=COUP%C3%89,%20Pierrick&MANSENCAL,%20Boris&CL%C3%89MENT,%20Micha%C3%ABl&GIRAUD,%20R%C3%A9mi&DENIS%20DE%20SENNEVILLE,%20Baudouin&rft.genre=article


Fichier(s) constituant ce document

FichiersTailleFormatVue

Il n'y a pas de fichiers associés à ce document.

Ce document figure dans la(les) collection(s) suivante(s)

Afficher la notice abrégée