基于EM算法的低剂量CT图像去噪

摘要

提高低剂量Cr图像的信噪比是使其获得有效临床应用的关键.文中针对低剂量CT投影数据极低信噪比特性以及投影数据噪声所特有的非平稳高斯特性,提出采用EM(Expectation-Maximization)算法通过求解图像后验概率的条件期望值最大的方法达到图像复原目的,同时在算法中实现了图像模型参数的估计,并且引入Gibbs采样技术,很好的解决了算法计算问题.计算机仿真及真实投影数据的实验表明,本文算法无论从复原图像的可视化效果上还是从噪声-分辨率关系的定量分析上,都具有一定优势.%Improving of the SNR of the low-dose CT image is a crucial issue for the low-dose CT application. In this paper, we employed an EM (expectation-maximization) scheme to restore the sinogram by the maximum a conditional expectation of the posteriori estimation,based on the special statistical property of low-dose CT sinogram, I. E., the extremely low SNR and the non-stationary noise property of the sinogram data. At the same time, parameters of the statistical model were estimated in the EM scheme.In addition,a Gibbs sampler was used to solve the computation problem.The effectiveness of the proposed algorithm was validated by both computer simulations and experimental studies.The gain of the proposed approach over other methods was quantified by noise-resolution tradeoff curves.