Context Aware 3D CNNs for Brain Tumor Segmentation

Abstract : In this work we propose a novel deep learning based pipeline for the task of brain tumor segmentation. Our pipeline consists of three primary components: (i) a preprocessing stage that exploits histogram standardization to mitigate inaccuracies in measured brain modalities, (ii) a first prediction stage that uses the V-Net deep learning architecture to output dense, per voxel class probabilities, and (iii) a prediction refinement stage that uses a Conditional Random Field (CRF) with a bilateral filtering objective for better context awareness. Additionally, we compare the V-Net architecture with a custom 3D Residual Network architecture, trained on a multi-view strategy, and our ablation experiments indicate that V-Net outperforms the 3D ResNet-18 with all bells and whistles, while fully connected CRFs as post processing, boost the performance of both networks. We report competitive results on the BraTS 2018 validation and test set.
Type de document :
Communication dans un congrès
MICCAI Brainlesion Workshop, Sep 2018, Granada, Spain
Liste complète des métadonnées

https://hal.inria.fr/hal-01959610
Contributeur : Maria Vakalopoulou <>
Soumis le : mardi 18 décembre 2018 - 19:04:49
Dernière modification le : jeudi 7 février 2019 - 17:29:13

Fichier

chandra_etal.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01959610, version 1

Citation

Siddhartha Chandra, Maria Vakalopoulou, Lucas Fidon, Enzo Battistella, Théo Estienne, et al.. Context Aware 3D CNNs for Brain Tumor Segmentation. MICCAI Brainlesion Workshop, Sep 2018, Granada, Spain. 〈hal-01959610〉

Partager

Métriques

Consultations de la notice

71

Téléchargements de fichiers

93