Anatomical and Functional Characterization in Children With Unilateral Cerebral Palsy: An Atlas-Based Analysis

Abstract

Background. Variability in hand function among children with unilateral cerebral palsy (UCP) might reflect the type of brain injury and resulting anatomical sequelae. Objective. We used atlas-based analysis of structural images to determine whether children with periventricular (PV) versus middle cerebral artery (MCA) injuries might exhibit unique anatomical characteristics that account for differences in hand function. Methods. Forty children with UCP underwent structural brain imaging using 3-T magnetic resonance imaging. Brain lesions were classified as PV or MCA. A group of 40 typically developing (TD) children served as comparison controls. Whole brains were parcellated into 198 structures (regions of interest) to obtain volume estimates. Dexterity and bimanual hand function were assessed. Unbiased, differential expression analysis was performed to determine volumetric differences between PV and MCA groups. Principal component analysis (PCA) was performed and the top 3 components were extracted to perform regression on hand function. Results. Children with PV had significantly better hand function than children with MCA. Multidimensional scaling analysis of volumetric data revealed separate clustering of children with MCA, PV, and TD children. PCA extracted anatomical components that comprised the 2 types of brain injury. In the MCA group, reductions of volume were concentrated in sensorimotor structures of the injured hemisphere. Models using PCA predicted hand function with greater accuracy than models based on qualitative brain injury type. Conclusions. Our results highlight unique quantitative differences in children with UCP that also predict differences in hand function. The systematic discrimination between groups found in our study reveals future questions about the potential prognostic utility of this approach.