Application of regularized least-squares algorithm in PET image reconstruction via a new anisotropic diffusion

The traditional iterative reconstruction algorithms of positron emission tomography cannot effectively suppress the noise. In order to solve the problem, a new anisotropic diffusion term are introduced into the least-squares algorithm, and with median filter the regularized least-squares algorithm in PET image reconstruction based on anisotropic diffusion come into being. Results of computer simulated demonstrate that compared with the other classical reconstruction algorithms, LS_NewAD not only availably suppress the noise and reconstruct a higher quality image, but also Effectively retains the edge of the image edge structure.