FIGURE 2.

The construction of individual‐level brain morphological similarity networks. (A) The basis for constructing individual‐level morphological brain networks: morphological metrics extracted from individual structural MRI. (B) The single metric‐based methods for constructing individual morphological similarity networks, such as calculating the statistical correlation of morphological metric between different cubes ²⁰ (top‐left panel), estimating the similarity in the distribution of regional morphological indicator ²¹ , ²² , ²⁷ (top‐right panel), and the network template perturbation approach ²⁹ (bottom‐left panel). (C) The multiple metrics‐based methods for constructing morphological similarity networks, including (1) defining multivariate Euclidean distance to depict multiple metrics‐based interregional similarity, ³¹ calculating the statistical correlation of (2) multiple morphological features (extracted from single‐modal or multimodal MRI) ³⁰ , ³² or (3) radiomics features between regions. ³³ (D) Each of the above methods ultimately generates the similarity matrix, which is subsequently used to generate the network graphs for further graph‐theoretic analyses. CT, cortical thickness; FA, fractional anisotropy; FI, folding index; GC, Gaussian curvature; GM, gray matter; GMV, gray matter volume; IC, intrinsic curvature; MC, mean curvature; MD, mean diffusivity; MRI, magnetic resonance imaging; MT, magnetization transfer; PD, probability density; PDF, probability density function; SA, surface area; SD, sulcal depth; WM, white matter.