A hierarchical independent component analysis model for longitudinal neuroimaging studies

Area of Science:

• Neuroscience
• Biostatistics
• Medical Imaging Analysis

Background:

• Longitudinal neuroimaging studies are crucial for understanding brain changes over time in disease, development, and treatment.
• Independent Component Analysis (ICA) is widely used for brain network analysis but is limited to cross-sectional data.
• Existing methods cannot effectively model the temporal dynamics of brain networks in longitudinal studies.

Purpose of the Study:

• To develop a novel longitudinal independent component model (L-ICA) for analyzing brain functional networks in longitudinal neuroimaging data.
• To provide a formal framework for extending ICA to accommodate repeated measurements and subject-specific effects.
• To enable accurate estimation of population- and individual-level changes in brain networks and their modulation by covariates.

Main Methods:

• Proposed a longitudinal independent component model (L-ICA) incorporating subject-specific random effects and visit-specific covariate effects.
• Developed an exact EM algorithm for maximum likelihood estimation and a subspace-based approximate EM algorithm for computational efficiency.
• Introduced a statistical testing procedure to examine covariate effects on brain network changes.

Main Results:

• L-ICA provides more accurate estimates of brain network changes over time compared to existing cross-sectional methods.
• The model effectively borrows information across repeated scans, increasing statistical power to detect covariate effects.
• Application to the ADNI2 study revealed novel, biologically insightful findings in Alzheimer's disease brain network changes.

Conclusions:

• L-ICA offers a powerful and flexible framework for analyzing longitudinal neuroimaging data, advancing the study of dynamic brain functional networks.
• The developed algorithms provide accurate and computationally efficient solutions for L-ICA.
• L-ICA has significant potential for investigating disease progression, treatment effects, and neurodevelopmental changes in brain networks.