基于变分光流估计的肺部4D-CT图像超分辨率重建

发布时间：2018-04-09 05:21

本文选题：D-CT图像　切入点：超分辨率重建　出处：《计算机研究与发展》2017年08期

【摘要】：由于受到扫描时间和照射剂量的限制,肺部4D-CT数据中纵向采样率远小于面内采样率.为了得到更高质量的肺部图像,从医学图像固有的自相似性出发,提出了一种基于局部和全局相结合的变分光流估计的图像序列超分辨率重建技术,用于提高4D-CT图像重建质量.首先,构建了一个用于求解肺部4D-CT不同相位图像之间的光流场的变分光流模型;然后,利用快速交替方向乘子法求解该模型,得到不同相位图像之间的光流场;最后,基于光流场,并利用非局部迭代反投影超分辨率重建算法,实现了高分辨率肺部图像的重建.实验结果表明:与已有算法相比,本方法在增强图像纹理结构的同时更好地保留了图像的轮廓.
[Abstract]:Due to the scanning time and radiation dose limit, lung 4D-CT data in longitudinal sampling rate is far less than the in-plane sampling rate. In order to get the lung images with higher quality, from the self similarity of natural medical image, this paper proposes a local and global based on the combination of variational optical flow estimation of image sequence super resolution reconstruction technique for improving 4D-CT image reconstruction quality. First, one for solving between lung 4D-CT images of different phase optical flow field variational optical flow model is constructed; then, using fast alternating direction method of multipliers to solve the model, get the optical flow between different phase image; finally, based on the optical flow field, and the use of non local iterative back projection super-resolution reconstruction algorithm, to realize the reconstruction of high resolution lung images. Experimental results show that compared with the existing algorithms, this method in the enhancement of image texture and structure The outline of the image is preserved well.

【作者单位】：山东财经大学计算机科学与技术学院;山东省数字媒体技术重点实验室;山东大学计算机科学与技术学院;
【基金】：国家自然科学基金项目(61572286,61472220) 山东省重点研发项目(2014GGX101037) 济南市高校自主创新项目(201401216) 山东省高校优势学科人才团队培育项目~~
【分类号】：R734.2;TP391.41

【参考文献】