Leveraging 3D information in unsupervised brain MRI segmentation

0 avis

Leveraging 3D information in unsupervised brain MRI segmentation

Archive ouverte

Edité par CCSD ; IEEE -

International audience. Automatic segmentation of brain abnormalities is challenging, as they vary considerably from one pathology to another. Current methods are supervised and require numerous annotated images for each pathology, a strenuous task. To tackle anatomical variability, Unsupervised Anomaly Detection (UAD) methods are proposed, detecting anomalies as outliers of a healthy model learned using a Variational Autoencoder (VAE). Previous work on UAD adopted 2D approaches, meaning that MRIs are processed as a collection of independent slices. Yet, it does not fully exploit the spatial information contained in MRI. Here, we propose to perform UAD in a 3D fashion and compare 2D and 3D VAEs. As a side contribution, we present a new loss function guarantying a robust training. Learning is performed using a multicentric dataset of healthy brain MRIs, and segmentation performances are estimated on White-Matter Hyperintensities and tumors lesions. Experiments demonstrate the interest of 3D methods which outperform their 2D counterparts.

Langue: en

Indexation

Consulter en ligne

Consulter le document

Suggestions

Du même auteur

Uncertainty-based Quality Control for Subcortical Structures Segmentation in T1-weighted Brain MRI

Archive ouverte | Lambert, Benjamin | CCSD

International audience. Deep Learning (DL) models are presently the gold standard for medical image segmentation. However, their performance may drastically drop in the presence of characteristics in test images not...

Safety-Net: Automatic identification of segmentation errors for Multiple Sclerosis lesions. Safety-Net : Identification automatique des erreurs de segmentation des lésions de la Sclérose-en-Plaques

Archive ouverte | Lambert, Benjamin | CCSD

International audience. In the context of Multiple Sclerosis (MS), neural networks have the potential to greatly assist neurologists in personalizing therapy through automatic quantification of the evolution of brai...

Trustworthy clinical AI solutions: a unified review of uncertainty quantification in deep learning models for medical image analysis

Archive ouverte | Lambert, Benjamin | CCSD

The full acceptance of Deep Learning (DL) models in the clinical field is rather low with respect to the quantity of high-performing solutions reported in the literature. Particularly, end users are reluctant to rely on the rough ...

Chargement des enrichissements...