MRAU-net网络下的X光胸片肺野分割算法  

Lung Field Segmentation Algorithm of X-Ray Chest Film Based on MRAU-Net Network

作　　者：胡俊 李平[1] HU Jun;LI Ping(College of Information Science and Engineering,Huaqiao University,Xiamen 361021,China)

机构地区：[1]华侨大学信息科学与工程学院,福建厦门361021

出　　处：《华侨大学学报（自然科学版）》2023年第3期398-406,共9页Journal of Huaqiao University(Natural Science)

基　　金：国家自然科学基金资助项目(61603144);福建省自然科学基金资助项目(2018J01095);福建省高校产学合作科技重大项目(2013H6016);华侨大学中青年教师科技创新资助计划项目(ZQN-PY509)。

摘　　要：为了解决U-net网络进行X光胸片肺野分割时,受限于特征提取能力不足导致分割结果不精确的问题,提出一种多尺度残差注意力U型网络(MRAU-net)模型.利用多尺度信息融合(MIF)模块,改善网络结构,增加对多尺度信息的获取;利用通道和空间双注意力(CSDA)模块,解决网络在有限算力下的信息过载问题.同时,对残差模块进行改进,并与U-net网络进行深度结合,提升网络的学习稳定性,缓解梯度消失和过拟合现象.实验结果表明:文中方法具有优秀的X光胸片肺野分割能力,能获得更精确的分割结果.In order to solve the problem of imprecise segmentation results caused by insufficient feature extraction ability when U-net network is used to segment lung fields in X-ray chest films,a multi-scale residual attention U-net(MRAU-net)model is proposed.The multi-scale information fusion(MIF)module is used to improve the network structure and increase the acquisition of multi-scale information.Using channel and space dual attention(CSDA)module,the problem of information overload in the network under limited computing power is solved.At the same time,the residual module is improved and deeply combined with the U-net network to improve the learning stability of the network and alleviate the phenomenon of gradient disappearance and over fitting.The experimental results show that the proposed method has excellent segmentation ability of lung field in X-ray chest film,and can obtain more accurate segmentation results.

关 键 词：胸片肺野分割 U-net网络 多尺度信息融合模块 通道和空间双注意力模块 深度残差 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] R816.41[自动化与计算机技术—计算机科学与技术]