Fig. 3

The weights of the brain regions in age prediction within the TDC sample, using a all five metrics (volume, surface area, thickness, FA, and ADC) and b only FA. The second model is included here as it is the only model with a significant correlation between the DDI and ASD symptom severity. Only top 30 features are visualized. Most features were related to FA and ADC metrics in the combined model. Higher absolute values indicate bigger contribution in the prediction. Note that the weights should not be compared between the two models, as the difference in magnitudes can be mostly explained by the number of features used in models. The abbreviations used: L, left; R, right; Sub, subcortical; GM, gray matter; WM, white matter; Snigra, substantia nigra; ICing, isthmus of cingulate; CP, cerebral peduncle; RedNc, red nucleus; ENT, entorhinal; LOF, lateral orbitofrontal; MOF, medial orbitofrontal; ML, medial lemniscus; MF, middle frontal; SM, supramarginal; GCC, genu of corpus callosum; CGC, cingulum (cingulate gyrus); PrCe, precentral; PrCu, precuneus; MCP, middle cerebellar peduncle; UNC, uncinate; ST, superior temporal; LFO, lateral fronto-orbital; LFOG, lateral fronto-orbital gyrus; PHG, parahippocampal gyrus; MO, middle occipital; ACR, anterior corona radiata; GP, globus pallidus; STG, superior temporal gyrus; SS, sagittal stratum; MFOG, middle fronto-orbital gyrus; SCR, superior corona radiata; PrCG, precentral gyrus