基于CAMU-Net的肺结节分割方法  

Segmentation of lung nodules based on CAM U-Net

作　　者：王统 徐胜舟[1,2] 卢浩然 吴福彬 裴承丹 WANG Tong;XU Shengzhou;LU Haoran;WU Fubin;PEI Chengdan(College of Computer Science,South-Central Minzu University,Wuhan 430074,China;Hubei Provincial Engineering Research Center for Intelligent Management of Manufacturing Enterprises,Wuhan 430074,China;Network Information Center,Wuhan Institute of Technology,Wuhan 430205,China)

机构地区：[1]中南民族大学计算机科学学院,武汉430074 [2]湖北省制造企业智能管理工程技术研究中心,武汉430074 [3]武汉工程大学网络信息中心,武汉430205

出　　处：《中南民族大学学报（自然科学版）》2024年第1期104-111,共8页Journal of South-Central University for Nationalities：Natural Science Edition

基　　金：湖北省自然科学基金资助项目(2020CFB541);中央高校基本科研业务费专项资金项目(CZY19011)。

摘　　要：肺癌作为世界上死亡率最高的癌症之一,严重威胁人类的生命安全,早发现早治疗可以提高患者的生存率.为了准确地分割出肺部CT图像中的肺结节区域,提出一种基于CAM U-Net的肺结节分割方法.在U-Net网络基础上,通过添加通道注意力模块CAM,使网络中的特征聚焦于关键有用的信息,减弱甚至消除无关信息的干扰,进而提升模型的性能.在LIDC-IDRI肺结节公开数据集上的实验结果表明:该算法的交并比、Dice相似系数、准确率、和召回率分别为82.04%、89.24%、88.61%和91.28%.与其他肺结节分割方法相比,该算法具有更好的分割性能.As one of the cancers with the highest mortality rate in the world,lung cancer is a serious threat to human life.Early detection and early treatment can improve the survival rate of patients.In order to accurately segment the pulmonary nodules in the pulmonary CT images,a CAM U-Net based pulmonary nodules segmentation method was proposed.On the basis of U-Net network,the channel attention module CAM is added to make the features in the network focus on the key useful information,weaken or even eliminate the interference of irrelevant information,so as to improve the performance of the model.Experimental results on the LIDC-IDRI public dataset of pulmonary nodules show that the proposed method can achieve intersection over union,Dice similarity coefficient,precision and recall of 82.04%,89.24%,88.61%and 91.28%,respectively.Compared with other segmentation methods for lung nodules,this algorithm has better segmentation performance.

关 键 词：肺结节 分割 U-Net网络 通道注意力模块 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]