volumetric analysis that includes a library of MRI scans that has been used in prior studies (Frazier et al. 2005a; Frazier et al. 2005b; Frazier et al. 2008).

Healthy controls had no DSM-IV Axis I diagnosis based on structured and clinical interviews. Healthy controls had no first-degree family history of BP, ADHD, psychosis, or any other psychiatric family history

Pearson and Spearman (rank) correlations were performed on clinical variables (bipolar onset and duration, ADHD onset and duration, current GAF, and MRS mania and psychosis scores), the number of psychoactive medications (atypicals, stimulants, mood stabilizers, lithium, antidepressants, chlorpromazine equivalents), and age for only those structures that differed significantly between diagnostic groups and HC

Post hoc mean comparisons were made for significant main effects and interactions using Tukey's Honestly Significant Difference, with α = 0.05 to control for pairwise comparisons, and by the Student t-test to indicate modest (uncorrected) effects

SPSS 15.0 for Windows (SPSS, Inc., Chicago, IL) was used for statistical analysis

analyses of variance for continuous variables and chi-squared tests for categorical variables.

For each set of regions, a general linear mixed model with an unstructured covariance matrix was run to estimate overall diagnosis effects while controlling the multicolinearity among the regions of interest. Given a significant region-by-diagnosis effect, a series of univariate analyses of covariance (ANCOVA) were performed with diagnosis (HC, ADHD, BP + ADHD, BP ) and sex (female, male), plus their interaction, as factors, and with age and TCV as covariates. TCV was excluded as a covariate in the analysis of symmetry coefficients (age was the only covariate)

Each dataset was segmented according to the anatomic boundaries described in detail in Filipek et al. (1994) and Frazier et al. (2005a, 2005b). In brief, structural scans were positionally normalized to overcome variations in head position and then segmented into gray, white, and cerebrospinal fluid (CSF) tissue classes. The segmentation method uses a semiautomated intensity contour algorithm for external border definition and signal intensity histogram distributions for delineation of gray-white borders

Exclusion criteria for all subjects and HC were: major sensorimotor handicaps (e.g., deafness, blindness, paralysis); full scale intelligence quotient (IQ) < 70 or learning disabilities; history of claustrophobia, autism, schizophrenia, anorexia nervosa or bulimia, drug or alcohol dependence/abuse (during 2 months prior to scan or total past history ≥12 months); active medical or neurological disease; history of electroconvulsive therapy (ECT); metal fragments or implants; and current pregnancy or lactation. History of learning disabilities was obtained via parental interview, and these youths were excluded due to the potential for confounding of neuroanatomical findings

Inclusion criteria for all subjects in this analysis were: age 6–19 years old, right-handedness