Eye-gazeGuidedMulti-modalAlignmentforMedical RepresentationLearning