Comparison of denoising results. The averaged low-dose gated image generated from different motion estimation methods are shown in the 1st row. The corresponding denoised images are shown in the 2rd row. From left to right: ground truth, UNet denoising from the averaged image without any deformation, UNet denoising on the averaged image based on NRB-derived deformation fields, UNet denoising on the averaged image based on VM-derived deformation fields, UNet denoising on the averaged image based on SAN-derived deformation fields, and the end-to-end output from our MDPET. Our MDPET can reduce the motion blurring between the liver and kidney (gray box), as well as improving the visualization of small anatomic structures, such as portal veins (blue arrows).