V-net: Fully convolutional neural networks for volumetric medical image segmentation

F Milletari, N Navab, SA Ahmadi

Abstract: Convolutional Neural Networks (CNNs) have been recentlyemployed to solve problems from both the computer vision and medi-cal image analysis fields. Despite their popularity, most approaches areonly able to process 2D images while most medical data used in clinicalpractice consists of 3D volumes. In this work we propose an approachto 3D image segmentation based on a volumetric, fully convolutional,neural network. Our CNN is trained end-to-end on MRI volumes depict-ing prostate, and learns to predict segmentation for the whole volume atonce. We introduce a novel objective function, that we optimise duringtraining, based on Dice coefficient. In this way we can deal with situa-tions where there is a strong imbalance between the number of foregroundand background voxels. To cope with the limited number of annotatedvolumes available for training, we augment the data applying randomnon-linear transformations and histogram matching. We show in our ex-perimental evaluation that our approach achieves good performances onchallenging test data while requiring only a fraction of the processingtime needed by other previous method.