《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (9): 2940-2947.DOI: 10.11772/j.issn.1001-9081.2023030381

• 多媒体计算与计算机仿真 • 上一篇    下一篇

基于空洞卷积的医学图像超分辨率重建算法

李众1,2, 王雅婧1, 马巧梅1,2()   

  1. 1.中北大学 软件学院,太原 030051
    2.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:2023-04-07 修回日期:2023-07-03 接受日期:2023-07-06 发布日期:2023-09-10 出版日期:2023-09-10
  • 通讯作者: 马巧梅
  • 作者简介:李众(1974—),男,山西忻州人,副教授,博士,CCF会员,主要研究方向:图形图像处理、算法分析
    王雅婧(1998—),女,山西晋中人,硕士研究生,主要研究方向:图像处理、算法分析;
  • 基金资助:
    山西省自然科学基金资助项目(20210302123019)

Super-resolution reconstruction algorithm of medical images based on dilated convolution

Zhong LI1,2, Yajing WANG1, Qiaomei MA1,2()   

  1. 1.School of Software,North University of China,Taiyuan Shanxi 030051,China
    2.Shanxi Medical Imaging Artificial Intelligence Engineering Technology Research Center (North University of China),Taiyuan Shanxi 030051,China
  • Received:2023-04-07 Revised:2023-07-03 Accepted:2023-07-06 Online:2023-09-10 Published:2023-09-10
  • Contact: Qiaomei MA
  • About author:LI Zhong, born in 1974, Ph. D., associate professor. His research interests include graphics and image processing, algorithm analysis.
    WANG Yajing, born in 1998, M. S. candidate. Her research interests include image processing, algorithm analysis.
  • Supported by:
    Natural Science Foundation of Shanxi Province(20210302123019)

摘要:

为解决现有医学图像超分辨率重建中存在的图像细节模糊、全局信息利用不充分等问题,提出一种基于空洞卷积与改进的混合注意力机制的医学图像超分辨率重建算法。首先,将深度可分离卷积与空洞卷积相结合,使用不同大小的感受野对图像进行不同尺度的特征提取,从而增强特征表达能力;其次,引入边缘通道注意力机制,在提取图像高频特征的同时融合边缘信息,从而提高模型的重建精度;再次,混合L1损失与感知损失函数作为整体损失函数,使重建后的图像效果更符合人类视觉感观。实验结果表明,在放大因子为3时,与基于卷积神经网络的图像超分辨率(SRCNN)算法、VDSR(Very Deep convolutional networks Super-Resolution)相比,所提算法的峰值信噪比(PSNR)平均提高了11.29%与7.85%;结构相似性(SSIM)平均提高了5.25%和2.44%。可见,所提算法能增强医学图像的效果与纹理特征,且对图像整体结构还原更加完整。

关键词: 超分辨率重建, 医学图像, 深度可分离卷积, 空洞卷积, 注意力机制

Abstract:

To solve the problems such as blurred image details and insufficient utilization of global information in existing medical image super-resolution reconstruction, a medical image super-resolution reconstruction algorithm based on dilated convolution and improved hybrid attention mechanism was proposed. Firstly, depthwise separable convolution was combined with dilated convolution to extract image features at different scales by using perceptive fields with different sizes, thereby enhancing feature representation ability. Secondly, edge channel attention mechanism was introduced to fuse the edge information while extracting the high-frequency image features, thereby improving the reconstruction accuracy of the model. Thirdly, L1 loss and perceptual loss were mixed as the overall loss function in order to make the reconstructed image effect more consistent with human visual perception. Experimental results show that when the amplification factor is 3, compared with Super-Resolution Convolutional Neural Network (SRCNN) and VDSR (Very Deep convolutional networks Super-Resolution), the proposed algorithm has the PSNR (Peak Signal-to-Noise Ratio) improved by 11.29% and 7.85% averagely and respectively, and the SSIM (Structural Similarity Index Measure) improved by 5.25% and 2.44% averagely and respectively. It can be seen that the proposed algorithm enhances the effect and texture features of the medical images, and provides more complete restoration of overall image structure.

Key words: super-resolution reconstruction, medical image, depthwise separable convolution, dilated convolution, attention mechanism

中图分类号: