Multivoxel pattern analysis of structural MRI in children and adolescents with conduct disorder

Conduct disorder (CD) is a psychiatric disorder in either childhood or adolescence and is characterized by aggressive and antisocial behavior. Although CD has been shown to be associated with structural abnormalities by structural magnetic resonance imaging (sMRI), the classification ability of these structural abnormalities' spatial patterns remains unclear. The aim of the present study was to characterize these different spatial patterns, which may eventually serve as potential reliable imaging biomarkers in the classification of CD from healthy controls (HCs). High-resolution 3D sMRI was acquired from 60 CD and 60 HCs, and all subjects were male participants. The mean (standard deviation) age was 15.3 (1.0) years old and 15.5 (0.7) years old for the CD and HC group respectively. Multivoxel pattern analysis (MVPA) with searchlight algorithm combined with support vector machine (SVM) was used to characterize the different spatial patterns in grey matter (GM) and to assess the classification ability of such structural difference. Seven cortical and subcortical regions showed significant GM difference between CD and HCs, including the cerebellum posterior lobe, temporal lobe, parahippocampal gyrus, lingual gyrus, insula, parietal lobe and medial frontal gyrus. GM in these brain regions discriminated CD with accuracy of up to 83%. Multiple brain regions exhibited aberrantly different spatial patterns in CD. The spatial patterns might be objective and reliable imaging features that could help to improve the classification of CD.