低剂量CT图像恢复算法综述

Review on low-dose CT image restoration algorithms

本文主要针对低剂量计算机断层成像(Low-Dose Computed Tomography,LDCT)的重建图像中含有高噪声和伪影等问题,综述了近几年LDCT图像恢复算法的国内外进展。其处理方法大致可以分为三类:投影域预处理、迭代重建算法和图像后处理。投影域预处理方法利用含噪投影数据的统计特性估计无噪声的情况。迭代重建方法因其良好的抗噪性能和不断提高的计算能力而得到广泛应用。与前两类相比,图像后处理方法可以直接应用到重建后的CT图像上,并且很容易整合到当前的CT工作流中。文末还对LDCT图像恢复的发展进行了讨论。

In this paper,we discuss the progress of low-dose computed tomography(LDCT)image restoration algorithms during recent years,aiming at the problems of severe noise and artifacts in LDCT reconstructed images.The processing methods can be divided into three categories:projection preprocessing,iterative reconstruction algorithms,and image post-processing.Projection preprocessing utilizes the statistical characteristics of noisy projection data to estimate the clear data.Iterative reconstruction methods have been widely used because of anti-noise performance and the continuously improving computing power.Compared with the first two categories,image post-processing methods can be directly applied to reconstructed CT images and easily integrated into the current CT workflow.At the end of this paper,the development of LDCT image restoration is discussed.