计算机工程与应用 ›› 2017, Vol. 53 ›› Issue (15): 191-195.DOI: 10.3778/j.issn.1002-8331.1602-0198

• 图形图像处理 • 上一篇    下一篇

基于多相位图像相似性的肺4D-CT超分辨率重建

陈  瑾1,2,张  煜1,2   

  1. 1.南方医科大学 生物医学工程学院,广州 510515
    2.南方医科大学 广东省医学图像处理重点实验室,广州 510515
  • 出版日期:2017-08-01 发布日期:2017-08-14

Multi-phase similarity based super-resolution reconstruction for lung 4D-CT

CHEN Jin1,2, ZHANG Yu1,2   

  1. 1.School of Biomedical Engineering, Southern Medical University, Guangzhou 510515, China
    2.Guangdong Key Laboratory of Medical Image Processing, Southern Medical University, Guangzhou 510515, China
  • Online:2017-08-01 Published:2017-08-14

摘要: 肺4D-CT为放射治疗提供全面的图像引导,在当今肺癌治疗中起着重要的作用。然而,由于采集时间和人体所能承受辐射剂量的限制,无法得到高分辨率([Z]轴方向)CT图像,因此,通常采集到的肺4D-CT图像[Z]轴方向细节部分往往不够清晰。针对这一问题,提出了一个基于多相位相似性的非局部均值超分辨率重建方法,来提升肺4D-CT图像的质量。该方法利用多相位图像之间的互补信息,以非局部均值滤波为基础,来恢复图像的高分辨率细节结构;另外,由于呼吸运动,肺4D-CT图像中不同相位图像肺区域的灰度会有所差异,为保证重建高分辨率图像的灰度不变性,引入全局约束以修正重建图像的整体灰度。采用一套公共肺4D-CT数据集评估提出的方法,该数据集由10组肺4D-CT数据组成,每组数据包含10个相位。实验表明,在恢复图像的细节和增强分辨率方面,该方法要优于传统的线性插值和凸集投影(Projection Onto Convex Set,POCS)超分辨率重建算法。

关键词: 肺4D-CT, 非局部均值, 超分辨率重建

Abstract: Lung 4D-CT is of great value in the application of cancer treatment, because it provides comprehensive image guidance for radiation therapy. However, the high resolution lung 4D-CT data along the Z axis can’t be obtained due to the dose limitation. Concern this problem, a multi-phase similarity based non-local means super-resolution reconstruction for lung 4D-CT is proposed in this paper. The redundant information from multi-phases of 4D-CT data are used to recover some of this high frequency information by using a patch-based reconstruction in combination with a global constraint. A public dataset provided by DIR-lab is used to evaluate the proposed method. The dataset consists of 10 groups of lung 4D-CT data and each group contains 10 phases. All the evaluation results show that the approach outperforms traditional linear-interpolation and the convex set projection approach(POCS) in terms of quantitative measures and visual observation.

Key words: lung 4D-CT, non-local means, super-resolution reconstruction