A method may include: obtaining a 3D CT image of a scanning area of a subject; obtaining PET data of the scanning area of the subject; gating the PET data based on a plurality of motion phases; reconstructing a plurality of gated 3D PET images; registering the plurality of gated 3D PET images with a reference 3D PET image; determining a motion vector field corresponding to a gated 3D PET image of the plurality of gated 3D PET images based on the registration; determining a motion phase for each of the plurality of CT image layers; correcting, for each of the plurality of CT image layers, the CT image layer with respect to a reference motion phase; and reconstructing a gated PET image with respect to the reference motion phase based on the corrected CT image layers and the PET data.