Learning multi-scale synergic discriminative features for prostate image segmentation
作者：
Highlights：
• Propose multi-scale synergic discriminative network (MSD-Net) to segment the prostate gland in MR images.
• Construct two decoders to exploit semantically consistent features and intra-slice discriminative features separately.
• Design the CPC block and residual refinement block to fully exploit the multi-scale spatial contextual information.
• Introduce the synergic multi-task loss to impose the consistence constraint on the joint segmentation and boundary detection.
• Achieve the state-of-the-art performance on three public prostate MR image datasets.
摘要
•Propose multi-scale synergic discriminative network (MSD-Net) to segment the prostate gland in MR images.•Construct two decoders to exploit semantically consistent features and intra-slice discriminative features separately.•Design the CPC block and residual refinement block to fully exploit the multi-scale spatial contextual information.•Introduce the synergic multi-task loss to impose the consistence constraint on the joint segmentation and boundary detection.•Achieve the state-of-the-art performance on three public prostate MR image datasets.
论文关键词：Prostate segmentation,Intra-class consistency,Inter-class discrimination,Synergic multi-task loss
论文评审过程：Received 7 May 2021, Revised 17 January 2022, Accepted 24 January 2022, Available online 2 February 2022, Version of Record 7 February 2022.
论文官网地址：https://doi.org/10.1016/j.patcog.2022.108556
