低剂量CT图像全变分深度展开去噪网络  

Deep Total Variation Denoising Network for Low-Dose CT Images

作　　者：吴涵 张鹏程[1,2] 桂志国[1,2] 刘祎[1,2] WU Han;ZHANG Pengcheng;GUI Zhiguo;LIU Yi(College of Information and Communication Engineering,North University of China,Taiyuan 030051,China;Shanxi Key Laboratory of Biomedical Imaging and Big Data,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学信息与通信工程学院,太原030051 [2]中北大学生物医学成像与影像大数据山西重点实验室,太原030051

出　　处：《计算机测量与控制》2024年第12期229-235,共7页Computer Measurement &Control

基　　金：山西省基础研究计划项目(202303021211148,202103021224204,20210302124403);山西省回国留学人员科研资助项目(2021-111)。

摘　　要：对低剂量CT图像去噪进行了研究,分析了神经网络去噪在伪影抑制中计算性能低、泛化性不足的问题;采用各向异性全变分深度展开去噪网络,新方法结合图像相邻体素的边缘特性,引入各向异性TV正则项保留图像结构信息,避免各向同性TV导致的边缘模糊,并通过Chambolle-Pock算法求解数学模型,适配深度展开到卷积神经网络;此外,结合像素注意力机制进行网络优化,捕捉图像中的重要细节;经实验测试,基于Mayo 2016数据集,该方法在图像去噪效果上优于传统方法及其他先进网络模型,在PSNR、SSIM和VIF等指标上表现更优,满足低剂量CT图像高质量重建的需求。Research on denoising low-dose CT images was conducted,the issues of low computational performance and insufficient generalization in neural network denoising for artifact suppression were analyzed.An anisotropic total variation deep unfolding denoising network was adopted,with the new method incorporating the edge characteristics of adjacent voxels,the anisotropic TV regularization term was introduced to preserve the structural information of images and avoid the edge blurring caused by isotropic TV.The Chambolle-Pock algorithm was employed to solve the mathematical model,suitable for deep unfolding into convolutional neural networks.Additionally,the pixel attention mechanism was integrated for the network optimization to capture the important information of images.Through experimental tests on the Mayo 2016 dataset,this method has advantages over traditional methods and other advanced network models in image denoising,showing a better performance in the indicators of PSNR,SSIM,and VIF.This method meets the requirements for high-quality reconstruction of low-dose CT images.

关 键 词：图像去噪 计算机断层扫描 原始对偶算法 模型驱动 注意力机制 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]