A hierarchical independent component analysis model for longitudinal Neuroimaging studies

摘要: In recent years, longitudinal neuroimaging study has become increasingly popular in neuroscience research to investigate disease-related changes brain functions. current literature, one of the most commonly used tools extract and characterize functional networks is independent component analysis (ICA). However, existing ICA methods are not suited for modelling repeatedly measured imaging data. this paper, we propose a novel model (L-ICA) which provides formal modeling framework extending studies. By incorporating subject-specific random effects visit-specific covariate effects, L-ICA able provide more accurate estimates on both population- individual-level, borrow information across repeated scans within same subject increase statistical power detecting networks, allow model-based prediction caused by disease progression, treatment or neurodevelopment. We develop fully traceable exact EM algorithm obtain maximum likelihood L-ICA. further subspace-based approximate greatly reduce computation time while still retaining high accuracy. Moreover, present testing procedure examining network changes. Simulation results demonstrate advantages our proposed methods. apply ADNI2 Alzheimer disease. Results from biologically insightful findings revealed using