Analysis of a 3D Nonlinear Diffusion-based Filter with an Optimization Strategy Based on Low-dose MDCT Images

Abstract

For the problem that MDCT image quality is degraded by quantum noise heavily when the X-ray dose level is decreased, a 3D Nonlinear Diffusion-based Filter with an Optimization Strategy (NDF-OS) is analyzed and applied to improve these low dose images. The diffusion parameter of the 3D NDF-OS is decided according to the deduced dependency between the diffusion factor and gradients, instead of being set empirically. The other parameters are set according to the optimization strategy using a steepest descending method. 3D MDCT images with four dose levels are used to check the filtering performance. Experimental results show that the 3D NDF-OS works well and can achieve a better trade-off between smoothing efficiency and feature preservation than other diffusion methods.