基于稀疏采样的CT快速成像方法研究

Arapid CT imaging method based on sparse sampling

针对CT高分辨率成像的数据采集速率问题,本文提出了一种基于图像压缩感知原理的CT快速成像方法。对目标物体进行稀疏角度的投影采样,以压缩重构图像所需的投影量,进而提升图像采集速度。重建过程中,引用CT领域中的重建方案,为了在不完备的数据情况下保留图像的灰度信息,本文采用迭代法中的代数重建算法(ART)进行图像重构。由于稀疏采样下ART重建图像存在较严重的伪影,本文将全变分最小化约束加入到ART算法中,实现对目标物体的精确重建,并将重构结果分别与压缩采样匹配追踪算法(CoSaMP)、Split-Bregman算法(SpBr)进行比较。实验结果表明,本算法在低采样率下的MSSIM均高于CoSaMP算法和SpBr算法。能有效保留图像的灰度信息和提升图像采集速度。

Aiming at the problem of data acquisition rate for high-resolution CT imaging,this paper presents a rapid CT imaging method based on image compression sensing principle. The target object is sampled by sparse angle to compress the projection volume,thereby enhancing the image acquisition speed. In order to reconstruct the grayscale information of the object in the case of incomplete data,this paper uses the algebraic reconstruction algorithm（ ART） solved iteratively to reconstruct the image. Moreover,total variation constraint is added to the ART algorithm to improve the reconstruction. The reconstructed result is compared with the compression sampling matching pursuit（ CoSaMP） algorithm and Split-Bregman（ SpBr） algorithm respectively. The experimental results show that the MSSIM of this algorithm is higher than that of CoSaMP algorithm and SpBr algorithm at a low sampling rate. This method can effectively preserve the grayscale information of images and enhance the data acquisition speed.