《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (2): 584-591.DOI: 10.11772/j.issn.1001-9081.2021020219

• 多媒体计算与计算机仿真 • 上一篇    

基于超分辨率网络的CT三维重建算法

李俊伯1,2,3, 秦品乐1,2,3, 曾建潮1,2,3(), 李萌1,2,3   

  1. 1.山西省医学影像与数据分析工程研究中心(中北大学), 太原 030051
    2.中北大学 大数据学院, 太原 030051
    3.山西省医学影像人工智能工程技术研究中心(中北大学), 太原 030051
  • 收稿日期:2021-02-04 修回日期:2021-05-11 接受日期:2021-05-13 发布日期:2021-06-02 出版日期:2022-02-10
  • 通讯作者: 曾建潮
  • 作者简介:李俊伯(1996—),男,山西运城人,硕士研究生,主要研究方向:机器学习、计算机视觉、医学影像分析;
    秦品乐(1978—),男,山西长治人,教授,博士,CCF会员,主要研究方向:医学影像分析、大数据、机器视觉;
    曾建潮(1963—),男,山西太原人,教授,博士,CCF会员,主要研究方向:医学影像、复杂系统的维护决策;
    李萌(1995—),男,山东济宁人,硕士研究生,主要研究方向:机器学习、计算机视觉、医学影像分析。
  • 基金资助:
    山西省工程技术研究中心建设项目(201805D121008)

CT three-dimensional reconstruction algorithm based on super-resolution network

Junbo LI1,2,3, Pinle QIN1,2,3, Jianchao ZENG1,2,3(), Meng LI1,2,3   

  1. 1.Shanxi Medical Imaging and Data Analysis Engineering Research Center (North University of China),Taiyuan Shanxi 030051,China
    2.College of Data Science,North University of China,Taiyuan Shanxi 030051,China
    3.Shanxi Medical Imaging Artificial Intelligence Engineering Technology Research Center (North University of China),Taiyuan Shanxi 030051,China
  • Received:2021-02-04 Revised:2021-05-11 Accepted:2021-05-13 Online:2021-06-02 Published:2022-02-10
  • Contact: Jianchao ZENG
  • About author:LI Junbo, born in 1996, M. S. candidate. His research interests include machine learning, computer vision, medical image analysis.
    QIN Pinle, born in 1978, Ph. D., professor. His research interests include medical image, big data, machine vision.
    ZENG Jianchao, born in 1963, Ph. D., professor. His research interests include medical image, maintenance decision of complex system.
    LI Meng, born in 1995, M. S. candidate. His research interests include machine learning, computer vision, medical image analysis.
  • Supported by:
    Construction Project of Engineering Technology Research Center of Shanxi Province(201805D121008)

摘要:

计算机断层扫描(CT)三维重建技术通过上采样体数据来提高三维模型质量,减轻模型中的锯齿状边缘、条纹状伪影和不连续表面等现象,从而提高临床医学中疾病诊断的准确率。针对以往CT三维重建后模型仍然不够清晰的问题,提出一种基于超分辨率网络的CT三维重建算法。网络模型为具有双重损失的优化学习纵轴超分辨率重建网络(DLRNet),通过单轴超分辨率进行腹部CT三维重建。网络末端引入优化学习模块,且除计算基准图与超分辨率图像的损失外,还计算网络内部粗略重建图像与基准图的损失,这样一来,优化学习与双重损失能使网络产生更接近于基准图的结果。随后在特征提取模块引入空间特征金字塔池化和通道注意力机制,加权细化学习了不同粗细以及规模不一的血管组织的特征。最后使用动态生成卷积核组的方法进行上采样使得单一网络模型可应对不同缩放因子的上采样任务。实验结果表明,相较于通道注意力的方法RCAN(Residual Channel Attention Network),所提网络模型在2、3、4倍缩放因子下的峰值信噪比(PSNR)平均提高0.789 dB。可见所提网络模型有效提升了CT三维模型的质量,一定程度上恢复了血管组织的连续细节特征,同时具备了实用性。

关键词: 深度学习, 三维重建, 超分辨率, 计算机断层扫描, 优化学习

Abstract:

Computed Tomography (CT) three-dimensional reconstruction technique improves the quality of three-dimensional model by upsampling volume data, and reduces the jagged edges, streak artifacts and discontinuous surface in the model, so as to improve the accuracy of disease diagnosis in clinical medicine. A CT three-dimensional reconstruction algorithm based on super-resolution network was proposed to solve the problem that the model after CT three-dimensional reconstruction remains unclear enough in the past. The network model is a Double Loss Refinement Network (DLRNET), and the three-dimensional reconstruction of abdominal CT was performed by uniaxial super-resolution. The optimization learning module was introduced at the end of the network model, and besides the calculation of the loss between the baseline image and super-resolution image, the loss between the roughly reconstructed image in the network model and the baseline image was also calculated. In this way, with the force of optimization learning and double loss, the results closer to the baseline image were produced by the network. Then, spatial pyramid pooling and channel attention mechanism were introduced into the feature extraction module to learn the features of vascular tissues with different thickness degrees and scales. Finally, the upsampling method was used to dynamically generate the convolution kernel set, so that a single network model was able to complete the upsampling tasks with different scaling factors. Experimental results show that compared with Residual Channel Attention Network (RCAN), the proposed network model improves the Peak Signal-to-Noise Ratio (PSNR) by 0.789 dB on average under 2, 3, and 4 scaling factors, showing that the network model effectively improves the quality of CT three-dimensional model, recovers the continuous detail features of vascular tissues to some extent, and has practicability.

Key words: deep learning, three-dimensional reconstruction, super-resolution, Computed Tomography (CT), optimization learning

中图分类号: