HD-Net: Hybrid Discriminative Network for Prostate Segmentation in MR Images

Abstract

Efficient and accurate segmentation of prostate gland facilitates the prediction of the pathologic stage and treatment response. Recently, deep learning methods have been proposed to tackle this issue. However, the effectiveness of these methods is often limited by inadequate semantic discrimination and spatial context modeling. To address these issues, we propose the Hybrid Discriminative Network (HD-Net), which consists of a 3D segmentation decoder using channel attention block to generate semantically consistent volumetric features and an auxiliary 2D boundary decoder guiding the segmentation network to focus on the semantically discriminative intra-slice features. Meanwhile, we further design the pyramid convolution block and residual refinement block for HD-Net to fully exploit multi-scale spatial contextual information of the prostate gland. In addition, to reduce the information loss in propagation and fully fuse the multi-scale feature maps, we introduce inter-scale dense shortcuts for both decoders. We evaluated our model on the Prostate MR Image Segmentation 2012 (PROMISE12) challenge dataset and achieved a synthetic score of 90.34, setting a new state of the art.