White-matter functional topology: a neuromarker for classification and prediction in unmedicated depression

摘要

Step 1: Structural images were co-registered with preprocessed functional images. Step 2: The co-registered structural image was segmented into WM, GM, and CSF. Step 3: The group-level WM mask was then randomly separated into 128 anatomical nodes with an approximately identical size. Step 4: BOLD-fMRI signals in WM were then obtained and used to compute FC matrices between each pair of nodes using Pearson’s correlation. Step 5: WM functional connectomes were constructed across a series of sparsity from 0.1–0.3 (interval = 0.01). Step 6: The AUC values of topological properties (i.e., small-world topology and nodal topological properties) were then evaluated across a series of sparsity. Finally, the small-world topology was used as a feature to predict depressive severity and to distinguish the patients from HCs. Abbreviation: AUC, area under curve; BOLD-fMRI, blood-oxygen-level-dependent functional magnetic resonance imaging; CSF, cerebrospinal fluid; FC, functional connectivity; GM, gray matter; HC, healthy control; MDD, major depressive disorder; WM, white matter; SVM, support vector machine; SVR, support vector regression.