基于小波与剪切波的压缩感知纵向MR成像

发布时间：2018-03-08 18:18

本文选题：图像处理　切入点：磁共振成像　出处：《北京交通大学学报》2017年03期 　论文类型：期刊论文

【摘要】：针对传统二维小波变换不能最优表示MR图像和MR图像数据采集缓慢的问题,基于相似性与联合稀疏变换对纵向压缩感知磁共振成像(LCS-MRI)提出新的正则化模型与相应的重建算法(frsLCS-MRI).为了验证联合稀疏变换与算法frsLCS-MRI的有效性,利用两个医学数据集与基于小波变换和相似性的算法LACS-MRI作了比较.数值实验表明:frsLCS-MRI算法可重建更高精度和更高信噪比的MR图像.使用联合稀疏变换比使用小波变换或剪切波变换可更好地重建MR图像的细节信息,如边界、拐角、轮廓、二维奇异曲线等.此外,利用相似性先验信息可节省数据采集时间.
[Abstract]:Aiming at the problem that the traditional two-dimensional wavelet transform can not best represent Mr image and the slow acquisition of Mr image data, Based on similarity and joint sparse transform, a new regularization model and a corresponding reconstruction algorithm for longitudinal compression perceptual magnetic resonance imaging (LCS-MRI) are proposed. In order to verify the validity of the combined sparse transform and algorithm frsLCS-MRI, a new regularization model and a corresponding reconstruction algorithm are proposed. Two sets of medical data are compared with LACS-MRI algorithm based on wavelet transform and similarity. Numerical experiments show that the weighted frsLCS-MRI algorithm can reconstruct Mr images with higher accuracy and higher signal-to-noise ratio. The joint sparse transform is better than the wavelet transform. The details of Mr image can be reconstructed better by changing or shearing wave transform. Such as boundary, corner, contour, two-dimensional singular curve, etc. In addition, the time of data acquisition can be saved by using similarity prior information.
【作者单位】：北京交通大学理学院;湘潭大学信息工程学院;北京航空航天大学精密光机电一体化技术教育部重点实验室;延安市人民医院MRICT诊断科;北京数学与信息交叉学科中心;
【基金】：国家自然科学基金(61071144,61271012)~~
【分类号】：R445.2;TP391.41

【相似文献】