Image-domain Noise Reduction with Multiscale Decomposition and Anisotropic Diffusion

Noise reduction in x-ray computed tomography (CT) is a critical technique for improving the diagnostic quality and saving radiation dose. Here, we propose an image-domain multiscale decomposition and anisotropic diffusion based noise reduction method for clinical CT applications. As used in the projection domain, a practical multiscale decomposition of CT image is first carried out using isotropic diffusion partial differential equation (PDE) in the image domain. Then, the image-domain anisotropic diffusion is adopted to reduce noise in each scale. Finally, in order to compensate for the degradation of image sharpness, an edge compensation step is followed. The performance of the proposed image-domain method for noise reduction is experimentally evaluated and verified using the scan data of an anthropomorphic head phantom acquired by a CT scanner. The preliminary result shows that the proposed image-domain multiscale decomposition-based anisotropic diffusion performs very well in noise reduction.