Fig. 2: Regional heterogeneity of extreme negative GMV deviations in each disorder.

a–f, Workflow for characterizing regional-level heterogeneity. GMV maps for each individual were parcellated into 1,000 cortical and 32 subcortical regions (a). The training dataset, HC_train, was used to train a normative model to make predictions about regional GMV values given an individual’s age, sex and scan site (b). The predictions for held-out controls (HC_test) and cases were then compared with empirical GMV estimates. Model predictions for one region, showing individuals in the training set (HC_train; light blue) and the held-out control (HC_test; dark blue) and clinical groups (red). Solid and dashed lines indicate the 99th and 95th centiles, respectively (c). For each individual, deviations from model predictions were quantified as a deviation \(z\) map (d). This deviation map was then thresholded at \(z > |2.6|\) to identify extreme deviations (e). For the HC_test and each clinical group, we quantified the proportion of individuals showing an extreme deviation in a given brain region, yielding an extreme deviation overlap map (f). We subtracted the HC_test overlap map from each clinical group’s overlap map to obtain an overlap difference map (\(\Delta\) overlap map) for each clinical group and then evaluated the magnitude of this difference (for details, see Extended Data Fig. 3a–d) (g). Cortical and subcortical surface renderings showing regions with significantly greater overlap of extreme negative GMV deviations in cases compared with controls, as identified using group-based permutation tests (pink corresponds to \({P}_{{{\mathrm{uncorrected}}}} < 0.05\), red corresponds to \({P}_{{{\mathrm{FDR}}}} < 0.05\); two tailed, cases > controls) (h). Data used to generate this figure can be found in Supplementary Data 1 (Regional_neg_thr26).