Toward Perception-Inspired and Physics-Informed Computational Models for Medical Image Quality Assessment and Enhancement