A Nonparametric model for Brain Tumor Segmentation and Volumetry in Longitudinal MR Sequences

Esther Alberts (1) , Guillaume Charpiat (2) , Yuliya Tarabalka (3) , Thomas Huber (4) , Marc-André Weber (5) , Jan Bauer (4) , Claus Zimmer (4) , Bjoern H. Menze (1)

1 TUM - Technische Universität Munchen - Technical University Munich - Université Technique de Munich
2 TAO - Machine Learning and Optimisation
3 TITANE - Geometric Modeling of 3D Environments
4 MRI TUM - Klinikum rechts der Isar
5 Division of Medical and Biological Informatics [Heidelberg]

Brain tumor image segmentation and brain tumor growth assessment are inter-dependent and benet from a joint evaluation. Starting from a generative model for multimodal brain tumor segmentation, we make use of a nonparametric growth model that is implemented as a conditional random field (CRF) including directed links with infinite weight in order to incorporate growth and inclusion constraints, reflecting our prior belief on tumor occurrence in the dierent image modalities. In this study, we validate this model to obtain brain tumor segmentations and volumetry in longitudinal image data. Moreover, we extend the framework with a probabilistic approach for estimating the likelihood of disease progression, i.e. tumor regrowth, after therapy. We present experiments for longitudinal image sequences with T1, T1c, T2 and FLAIR images, acquired for ten patients with low and high grade gliomas.

Domaines: Traitement du signal et de l'image [eess.SP]

Esther Alberts, Guillaume Charpiat, Yuliya Tarabalka, Thomas Huber, Marc-André Weber, et al.. A Nonparametric model for Brain Tumor Segmentation and Volumetry in Longitudinal MR Sequences. MICCAI Brain Lesion Workshop, Oct 2015, Munich, Germany. ⟨hal-01205916⟩