The conventional gating approach creates a dilemma for systems with large field-of-view (FOV) such as the uExplorer system. The motion effect is localized with variable motion amplitude across the whole FOV. A conventional gating approach would increase the noise for regions without motion, or with small motion amplitude. Existing local gating approach that allows optimized locally adaptive gating numbers requires a 4D reconstruction, which is time-consuming for generating of a 3D image. Image-domain motion amplitude is first estimated using a population-based or image-based approach. In the population-based approach, the motion amplitude is determined empirically based on the anatomy, and in the image-based approach, image registration is used. The motion amplitude for each line-of-response (LOR) was estimated using the forward projection approach. For time-of-flight (TOF) data, a regular forward projection was applied and for non-TOF data, a maximum intensity projection was applied. The useful percentage of counts for each LOR or TOF-LOR was determined by dividing the maximum allowed motion blurring amplitude by the motion amplitude for each LOR. The locally-gated sinogram data was generated by gating each LOR with the window determined by the useful percentage of counts. Locally gated sinograms were generated using the proposed gating method. More useful events can be preserved using the proposed approach when compared with conventional gating. Among the reconstructed images, the image reconstructed with locally gated sinograms was able to reduce motion blurring without an unnecessary increase in noise in regions unaffected by motion. We have proposed a novel local gating approach that can achieve locally-optimized resolution and noise tradeoff. The method can be directly incorporated into existing reconstruction methods with minimal modifications and extra computation.