| 501 | 0 | 7 |
| 下载次数 | 被引频次 | 阅读次数 |
医学图像分割作为当前的研究热点之一，分割精度对于后续的医学诊断影响巨大。针对目前大多数医学图像分割技术无法充分利用并融合多尺度特征信息的缺陷，提出了一种融合通道注意力机制的改进U-Net3+图像分割算法。在U-Net3+的全局跳跃连接结构的基础上，设计一种新的通道注意力机制并将它嵌入到U-Net3+网络的解码路径中，帮助分割网络在拼接全局特征图时调整重要信息的训练权重从而高效融合全局特征信息。最后，在两种经典的医学图像分割数据集上将该模型进行对比评估，平均Dice系数分别达到了74.31%和77.16%，相比原本U-Net3+的Dice系数分别提高了3.01%和2.98%。实验结果表明改进后的网络模型有效提高了医学图像的分割精度。
Abstract：Medical image segmentation is one of the current research hotspots,and the segmentation accuracy significantly impacts the subsequent medical diagnosis. In this paper,we proposed an improved U-Net3+ image segmentation algorithm that incorporated a channel attention mechanism to address the shortcomings of most current medical image segmentation techniques that can not fully utilize and fuse multi-scale feature information.Based on the global jump connection structure of U-Net3+,a new channel attention mechanism was designed and embedded into the decoding path of the U-Net3+ network to help the segmentation network adjust the training weights of important information when stitching the global feature map to fuse the global feature information efficiently. Finally,the model was compared and evaluated on two classical medical image segmentation datasets,and the average Dice coefficients reached 74. 31% and 77. 16%,respectively,were 3. 01%and 2. 98% higher than the original U-Net3+ Dice coefficients. The experimental results show that the improved network model effectively improves the segmentation accuracy of medical images.
[1]GUO Z,LI X,HUANG H,et al.Deep learning-based image segmentation on multimodal medical imaging[J].IEEE Transactions on Radiation and Plasma Medical Sciences,2019(3):162-169.
[2]LITJENS G,KOOI T,BEJNORDI B E,et al.A survey on deep learning in medical image analysis[J].Computer Vision and Pattern Recognition,2017(42):60-88.
[3]王春林，胡健民，张晓丹，等.医学图像分割研究进展与展望[J].中国医学物理学杂志，2017,34(2):215-223.
[4]PHAM D L,XU C,PRINCE J L.Current methods in medical image segmentation[J].Annual Review of Biomedical Engineering,2000(2):315-337.
[5]YAZID H,BASAH S N,RAHIM S A,et al.Performance analysis of entropy thresholding for successful image segmentation[C].2016 International Conference on Electrical,Electronics and Optimization Techniques(ICEEOT),2022:6433-6450.
[6]CANNY J.A computational approach to edge detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1986(6):679-698.
[7]BALI A,SINGH N S.A review on the strategies and techniques of image segmentation[C].2015 Fifth International Conference on Advanced Computing and Communication Technologies,Haryana,India,2015:113-120.
[8]RONNEBERGER O,FISCHER P,BROX T.U-Net:convolutional networks for biomedical image segmentation[C].Medical Image Computing and Computer-Assisted Intervention (MICCAI),Springer,Cham,2015:234-241.
[9]SHELHAMER E,LONG J,DARRELL T.Fully convolutional networks for semantic segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017(39):640-651.
[10]ZHANG Z,LIU Q,WANG Y.Road extraction by deep residual U-Net[J].IEEE Geoscience and Remote Sensing Letters,2018,15(5):749-753.
[11]HE K,ZHANG X,REN S,et al.Deep residual learning for image recognition[C].IEEE Conference on Computer Vision and Pattern Recognition(CVPR),2016:770-778.
[12]XIA X,KULIS B.W-Net:A deep model for fully unsupervised image segmentation[J].Computer Vision and Pattern Recognition,2017.DOI:10.48550/ar Xiv.1711.08506
[13]ZHOU Z,SIDDIQUEE M M R,TAJBAKHSH N,et al.UNet++:A nested U-Net architecture for medical image segmentation[C].Deep Learn Med Image Anal Multimodal Learn Clin Decis Support,Springer,Cham,2018:3-11.
[14]HUANG H,LIN L,TONG R,et al.UNet3+:A full-scale connected UNet for medical image segmentation[C].ICASSP 2020-2020 IEEE International Conference on Acoustics,Speech and Signal Processing(ICASSP),Barcelona,Spain,2020:1055-1059.
[15]沈楠，段友祥，孙岐峰，等.基于UNet3+的伽马成像测井自动解释方法[J].测井技术，2022,46(3):283-293.
[16]李霜.基于改进UNet3+和欠采样的雷达信号识别研究[D].长春：吉林大学，2022.
[17]詹光莉，刘辉，杨路.基于改进注意力W-Net的工业烟尘图像分割[J].计算机集成制造系统，2023,29(2):628-637.
[18]陈弘扬，高敬阳，赵地，等.LFSCA-UNet：基于空间与通道注意力机制的肝纤维化区域分割网络[J].中国图象图形学报，2021,26(9):2121-2134.
[19]吴倩倩，周蕾蕾，陆小妍，等.基于多头自注意力机制与U-Net的增强CT图像肾脏小肿瘤自动分割研究[J].中国医学装备，2022,19(2):27-31.
[20]易三莉，王天伟，杨雪莲，等.基于改进U-Net的肺野分割算法[J].激光与光电子学进展，2022,59(2):175-183.
[21]钟经纬.PRA-UNet3：全尺度跳跃连接CT肝脏图像分割模型[J].软件导刊，2023,22(2):15-20.
[22]SHAN T,YAN J.SCA-Net:A spatial and channel attention network for medical image segmentation[J].IEEE Access,2021(9):160926-160937.
[23]孙博，沈耿霖，赵岳飞.基于深度注意力机制的多任务医学图像分割[J].计算机工程与应用，2021,57(4):147-154.
[24]陈帆，毛凯明，张洪斌，等.一种基于多尺度和多任务注意力的病理图像分割算法[J].计算机应用研究，2021,38(4):1074-1079.
[25]STAAL J,ABRAMOFF M D,NIEMEIJER M,et al.Ridge-based vessel segmentation in color images of the retina[J].IEEE Transactions on Medical Imaging,2004,23(4):501-509.
[26]COVID-19 CT segmentation dataset[EB/OL].(2020-04-02)[2020-09-13].http：∥medicalsegmentation.com/covid19/.
基本信息:
DOI：10.16389/j.cnki.cn42-1737/n.2024.03.006
中图分类号:R318;TP391.41
引用信息:
[1]张全鑫,叶曦,杨志红等.基于改进通道多头注意力机制的U-Net3+医学图像分割算法研究[J].江汉大学学报(自然科学版),2024,52(03):51-61.DOI:10.16389/j.cnki.cn42-1737/n.2024.03.006.
基金信息:
工业烟尘污染控制湖北省重点实验室开放课题（HBIK2022-08）