CT male pelvic organ segmentation using fully convolutional networks with boundary sensitive representation.
Medical Image Analysis(2019)
摘要
•Robust and accurate soft labels are assigned to voxels near the boundary based on both spatial cue and semantic cue.•A multi-label cross-entropy loss that uses soft labels and hard labels as supervision is to perform segmentation.•A localization model is designed to focus on candidate regions, which can contribute significantly to better performance.•The experimental results on a challenging CT dataset show that our method outperforms the state-of-the-art methods.
更多查看译文
关键词
Image segmentation,Fully convolutional network,Boundary sensitive,Male pelvic organ,Prostate cancer,CT
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络