南方医科大学学报

南方医科大学学报 ›› 2022, Vol. 42 ›› Issue (6): 849-859.doi: 10.12122/j.issn.1673-4254.2022.06.08

• • 上一篇    下一篇

基于噪声水平估计的低剂量螺旋CT投影数据恢复

和法伟，王永波，陶 熙，朱曼曼，洪梓璇，边兆英，马建华   

  1. 南方医科大学生物医学工程学院，广东 广州 510515；琶洲实验室，广东 广州510330
  • 出版日期:2022-06-20 发布日期:2022-06-27

Low-dose helical CT projection data restoration using noise estimation

HE Fawei, WANG Yongbo, TAO Xi, ZHU Manman, HONG Zixuan, BIAN Zhaoying, MA Jianhua   

  1. School of Biomedical Engineering, Southern Medical University, Guangzhou 510515, China; Pazhou Lab, Guangzhou, 510330, China
  • Online:2022-06-20 Published:2022-06-27

RichHTML

28

PDF

247

摘要： 目的 构建任意低剂量水平下螺旋CT投影数据恢复模型。方法 首先，利用噪声估计模块估计任意低剂量水平的投影数据噪声方差图；然后利用估计出的噪声方差图指导投影数据进行恢复，即投影数据恢复模块；最后，采用滤波反投影算法进行图像重建。其中，噪声估计和投影数据恢复模块采用三维小波残差密集群网络结构，并利用非对称损失和全变分正则化进行约束。为验证方法的有效性，利用不同恢复模型分别对1/10、1/15常规剂量CT图像进行恢复，并与现有CT图像恢复网络IRLNet、REDCNN、MWResNet进行对比实验。 结果 在定量指标对比方面，本文提出的螺旋CT投影数据恢复方法相比于其他图像域恢复方法，结构相似性提升5.79%~17.46%（P<0.05）；临床影像医师图像质量评分结果显示，评分比其他方法高7.19%~17.38%（P<0.05）。结论 本文提出的投影数据恢复模型能够有效抑制不同低剂量水平投影数据中的噪声和伪影，重建得到组织结构完整、高分辨率的CT图像。

关键词: 投影数据恢复；低剂量螺旋CT成像；深度学习；噪声估计

Abstract: Objective To build a helical CT projection data restoration model at random low-dose levels. Methods We used a noise estimation module to achieve noise estimation and obtained a low-dose projection noise variance map, which was used to guide projection data recovery by the projection data restoration module. A filtering back-projection algorithm (FBP) was finally used to reconstruct the images. The 3D wavelet group residual dense network (3DWGRDN) was adopted to build the network architecture of the noise estimation and projection data restoration module using asymmetric loss and total variational regularization. For validation of the model, 1/10 and 1/15 of normal dose helical CT images were restored using the proposed model and 3 other restoration models (IRLNet, REDCNN and MWResNet), and the results were visually and quantitatively compared. Results Quantitative comparisons of the restored images showed that the proposed helical CT projection data restoration model increased the structural similarity index by 5.79% to 17.46% compared with the other restoration algorithms (P<0.05). The image quality scores of the proposed method rated by clinical radiologists ranged from 7.19% to 17.38% , significantly higher than the other restoration algorithms (P<0.05). Conclusion The proposed method can effectively suppress noises and reduce artifacts in the projection data at different low-dose levels while preserving the integrity of the edges and fine details of the reconstructed CT images.

Key words: projection data restoration; low-dose helical computed tomography; deep learning; noise estimation