医学图像分割的研究进展  

Review on medical image segmentation methods

作　　者：黄仟甲 张恒 李奇轩[1] 曹德政 焦竹青 倪昕晔[2,3,4,5] HUANG Qianjia;ZHANG Heng;LI Qixuan;CAO Dezheng;JIAO Zhuqing;NI Xinye(School of Computer Science and Artificial Intelligence,Changzhou University,Changzhou 213164,China;Department of Radiotherapy,Changzhou Second People's Hospital,Nanjing Medical University,Changzhou 213003,China;Jiangsu Province Engineering Research Center of Medical Physics,Changzhou 213003,China;Medical Physics Research Center,Nanjing Medical University,Changzhou 213003,China;Key Laboratory of Medical Physics in Changzhou,Changzhou 213003,China)

机构地区：[1]常州大学计算机与人工智能学院,江苏常州213164 [2]南京医科大学附属常州第二人民医院放疗科,江苏常州213003 [3]江苏省医学物理工程研究中心,江苏常州213003 [4]南京医科大学医学物理研究中心,江苏常州213003 [5]常州市医学物理重点实验室,江苏常州213003

出　　处：《中国医学物理学杂志》2024年第8期939-945,共7页Chinese Journal of Medical Physics

基　　金：国家自然科学基金(62371243);江苏省医学重点学科建设单位[肿瘤治疗学(放射治疗)](JSDW202237);江苏省重点研发计划社会发展项目(BE2022720);江苏省卫健委面上项目(M2020006);江苏省自然科学基金(BK20231190);常州市社会发展项目(CE20235063)。

摘　　要：医学图像是医生对患者进行病情诊断和治疗规划的有力工具。现今对于医学图像的分割不再局限于手工分割方法,通过传统方法与深度学习方法来实现医学图像分割已经取得更好、更准确的结果。本文基于近年来一些较为出众的医学图像创新分割方法进行综述,通过阐述深度学习方法如SAM、SegNet、MaskR-CNN和U-NET以及传统方法如活动轮廓模型、阈值分割模型创新等,对比各种图像分割方法的异同点,对医学图像分割方法做出总结与展望。以此来帮助学者们更好地了解目前的研究进展与未来的发展趋势。Medical image is a powerful tool to assist doctors in the diagnosis and treatment planning.Nowadays,the segmentation of medical images is no longer limited to manual segmentation methods.Traditional methods and deep learning methods have been used to achieve more accurate results in medical image segmentation.Herein some innovative medical image segmentation methods in recent years are reviewed.By elaborating on the innovations of deep learning methods(SAM,SegNet,Mask R-CNN,and U-NET)and traditional methods(active contour model and threshold segmentation model),the differences and similarities between them are compared.The summary of medical image segmentation methods and the prospect is expected to help researchers better grasp and familiarize themselves with research status and development trend.

关 键 词：医学图像分割 深度学习 阈值分割 神经网络 任意分割模型 综述 

分 类 号：R318[医药卫生—生物医学工程] TP391[医药卫生—基础医学]