(A–D) Independent validation of region-specific sexual dimorphism in C57BL/6J. Using data from an independent cohort of 663 males vs. 166 females, we display the volume distributions of regions found to display dimorphism in our basic cohort – medial amygdala (MEA) (A), bed nuclei of the stria terminalis (BST) (B), and ORBvl2/3 (C). Also shown is the similarity between whole-brain volumes (D). p-values are based on a rank-sum test. (E) The average p-value for sexual dimorphism in MEA (blue), BST (red), and ORBvl2/3 (blue) as a function of sample size. Specifically, n males and n females from our cohort were randomly selected and the rank-sum p-value was calculated. Shown is the average p-value over 1000 such random selections. (F) Top 10 regions when constructing a linear support vector machine (SVM) volume-based classifier. (G) Accuracy of volume-based classifiers when using the top 10 regions, adding one region at the time. (H, I) Same as (E, F) but for density-based classifiers. (J) Four examples of two-dimensional SVM based on volume information, as indicated on x and y axes; with accuracies indicated above. The separating line splits the plane into female/male domains. Red/blue markers represent individual males and females, respectively. Circles and diamonds represent correct and incorrect classifications, respectively.