A method and system for motion estimation and compensation are disclosed. Initially, a set of one or more initial images is reconstructed using acquired imaging data. Further, one or more regions of interest are identified in this set of reconstructed initial images. At least a set of filters is applied to the identified regions of interest to generate a sequence of filtered images. Particularly, each of the filtered images in the generated sequence of filtered images includes data acquired near a different reference point. Subsequently, a motion path corresponding to each region of interest is determined based on one or more correspondences in the sequence of filtered images.