Rethinking the Dice Loss for Deep Learning Lesion Segmentation in Medical Images
作者机构:Key Laboratory of Space Active Opto-Electronics TechnologyShanghai Institute of Technical PhysicsChinese Academy of SciencesShanghai 200083China University of Chinese Academy of SciencesBeijing 100049China Hangzhou Institute for Advanced StudyUniversity of Chinese Academy of SciencesHangzhou 310024China
出 版 物:《Journal of Shanghai Jiaotong university(Science)》 (上海交通大学学报(英文版))
年 卷 期:2021年第26卷第1期
页 面:93-102页
核心收录:
学科分类:1305[艺术学-设计学(可授艺术学、工学学位)] 0831[工学-生物医学工程(可授工学、理学、医学学位)] 13[艺术学] 08[工学] 081104[工学-模式识别与智能系统] 0804[工学-仪器科学与技术] 081101[工学-控制理论与控制工程] 0811[工学-控制科学与工程]
主 题:Dice loss deep learning medical image lesion segmentation
摘 要:Deep learning is widely used for lesion segmentation in medical images due to its breakthrough *** functions are critical in a deep learning pipeline,and they play important roles in segmenting *** loss is the most commonly used loss function in medical image segmentation,but it also has some *** this paper,we discuss the advantages and disadvantages of the Dice loss function,and group the extensions of the Dice loss according to its improved *** performances of some extensions are compared according to core *** different loss functions have different performances in different tasks,automatic loss function selection will be the potential direction in the future.