两种基于Anscombe变换域滤波的低剂量CT重建方法讨论

Comparison of Two Projection Restoration Based Low-dose CT Reconstruction Methods

目的:低剂量CT重建已成为CT成像的研究热点,针对低剂量CT成像质量退化问题,本文对两种基于Anscom-be变换域滤波的低剂量CT重建方法进行了讨论比较。方法:这两种方法分别为基于投影数据非单调性全变分最小恢复的低剂量CT重建方法和基于自适应模块匹配三维滤波低剂量CT重建方法。二者皆首先通过非线性Anscombe变换将满足Poisson分布的投影数据转化为近似Gaussian分布,其后再对变换后的Gaussian型数据分别进行非单调性全变分最小化算法(Nonmonotone Total Variation Minimization,NTVM)滤波和自适应模块匹配三维(Block-Matching and 3D,BM3D)滤波,最后再分别对Anscombe逆变换数据实现传统的滤波反投影(Filtered Back Projection,FBP)CT重建;我们通过数字体模仿真实验对两种方法分别进行了定性和定量比较分析。结果:实验结果表明,两种方法均能有效去除噪声、抑制伪影、提高低剂量CT重建图像质量。结论:BM3D滤波无需人工设置滤波参数,可实现自适应低剂量CT图像重建;重建时间方面,基于投影数据非单调性全变分最小恢复的低剂量CT重建方法更有优势。

Objetcive： Low-dose CT reconstruction has become the focus of CT imaging study. For low-dose CT imaging, two Anscombe transform based low-dose CT reconstruction methods were compared in this paper. Methods： These two reconstruction methods are based on projection restoration with non-monotone total variation minimization and Block-Matching and 3D, respectively. Projection data is transformed from Poisson distribution to Gaussian distribution using nonlinear Anscombe transform firstly. Then, the Anscombe transformed data is filtered by an efficient NTVM denoising algorithm and BM3D filtering, respectively. Last, the reconstruction is achieved by inverse Anscombe transform and FBP method, respectively. And finally the CT image was reconstructed by FBP approach. Simulated digital phantom experiments were conducted to compare these two methods qualitatively and quantitatively. Results： The results demonstrate that these two reconstruction methods can achieve a good performance in noise reduction and artifacts suppression. Conclusions： Because the variance of the Gaussian noise is known, the proposed scheme can be implemented without any manual parameters in BM3D. The method based on projection restoration with NTVM performs well in terms of reconstruction time.