v-net

A 3D fully convolutional neural network for volumetric medical image segmentation, where the loss function is directly defined on a Dice coefficient to deal with the class imbalance problem.