低剂量CT图像模型参数估计及统计去噪研究

Parameter estimation and statistical noise reduction for low-dose CT sinogram

由于提高低剂量CT图像的信噪比是低剂量CT获得有效临床应用的关键,为此,提出了一种低剂量CT投影域的自适应统计降噪算法.针对低剂量CT投影图像的非平稳高斯噪声特性,采用EM算法自适应地估计图像模型中的参数,并在此基础上对图像进行最大后验概率估计,从而达到图像降噪的目的.在对参数的估计过程中,引入MCMC的吉布斯采样技术,并在算法中引入两项初始化技术,从而减少了参数估计过程中的计算量,加快了算法的收敛速度.对仿真投影数据以及真实投影数据的实验结果表明,与传统算法相比,该算法在抑制噪声及保持分辨率方面均具有明显优势.

Improvement of the SNR of low-dose CT images is a crucial issue for the low-dose CT application. In this paper, we propose a novel adaptive statistical noise reduction algorithm for low-dose CT sinogram. The algorithm first adopts an EM algorithm to adaptively estimate the parameters of the image model based on the non-stationary Gaussian noise property in the low-dose CT projection data, and then uses the MAP estimation to restore the sinogram. In the parameters estimation procedure, a Gibbs sampler is used to handle the complicated computation problem. In addition, two initialization strategies are used in the algorithm to accelerate the convergence speed too. The effectiveness of the proposed algorithm is validated by both computer simulations and experimental studies. The advantage of the proposed approach over other methods is quantified by noise-resolution tradeoff curves.