Interpretation Scheme for Nonexpert Pediatricians Evaluating Magnetic Resonance Images of Children With Cerebral Palsy

This study evaluated the usability of our MRI interpretation scheme among pediatricians with different skill levels in evaluating MRI of patients with cerebral palsy. We divided MRI findings into three groups: no abnormalities, pre/perinatally acquired lesions, and other abnormalities. Pre/perinatally acquired lesions were divided into six subgroups. Other abnormalities included brain malformations, ventriculomegaly, atrophic changes, and other unclassifiable abnormalities. We compared the interpretations of eight participants, i.e., three nonexpert pediatricians, two junior pediatric neurologists, and three senior pediatric neurologists, in evaluating magnetic resonance images of 73 children with cerebral palsy. The degree of agreement was substantial or near perfect for all participants. When limited to pre/perinatally acquired lesions, the degree of agreement was near perfect for all but one participant. The rate of correct diagnosis did not differ greatly according to participants' experience with pre/perinatally acquired lesions. For patients with basal ganglia thalamic lesions, multicystic encephalomalacia, and posthemorrhagic porencephaly, the rate of correct diagnosis increased according to participants' experience. Pre/perinatally acquired lesions can be appropriately interpreted by nonexpert pediatricians utilizing our interpretation scheme.