基于参考图像的压缩感知磁共振扩散张量成像

Compressed sensing diffusion tensor imaging based on reference image

基于参考图像的压缩感知磁共振扩散张量成像方法,利用相邻方向的扩散加权图像差异较小的特点,采用压缩感知理论实现快速扩散张量成像,回顾性选取扩散张量图像数据进行实验研究,在采样率为50%的均匀分布辐射线欠采样方式下进行基于参考图像的压缩感知扩散张量图像重建,结果表明重建后的扩散加权图的平均结构相似性(MSSIM)和峰值信噪比(PSNR)分别为0.904±0.044、(37.92±3.89)dB,各向异性分数图的MSSIM和PSNR分别为0.992、41.64 dB。因此,该方法在保证重建图像质量的前提下,可显著缩短数据采集时间,减少由于时间过长引起的图像伪影等问题。

A method of compressed sensing diffusion tensor imaging based on reference image which utilizes the small differences between diffusion-weighted images in adjacent directions and compressed sensing theory to realize fast diffusion tensor imaging is proposed in the study,and diffusion tensor image data are retrospectively selected for experimental research.Compressed sensing diffusion tensor image reconstruction is performed based on reference image under a uniformly distributed radiation under-sampling method with a sampling rate of 50%.The results show that the mean structural similarity and peak signal-noise ratio of the reconstructed diffusion-weighted images are 0.904±0.044 and(37.92±3.89)dB,respectively,and those of fractional anisotropy images are 0.992 and 41.64 dB,respectively.Under the premise of ensuring the quality of the reconstructed image,using the proposed method can significantly shorten data acquisition time and reduce image artifacts caused by long imaging time.