Abstract: With the wide application of medical imaging technology,the low-dose CT has become an inevitable tendency in medical CT because of the gradual awareness of scan dose.However,the reconstruction image degradation in low-dose CT will affect the medical diagnostic accuracy.In order to solve the problem above,a4-neighborhood partial differential equation based on anisotropic diffusion equation sinogram noise reduction is proposed.To get more effective at noise smoothing and features preserving,novel model defined a weight normalized gradient according to the control rate of relevant pixel values.Additionally,a time-varied diffusion function based on the regular rule of image feature is provided to be adaptive for the balance between noise removal and edge preservation.Numerical results and scan data of the breast model reveal that the novel method can achieve better reconstruction image quality.