|
Abstract:
|
Currently, there has been strong interest in network-oriented research on brain functions and organizations. Independent component analysis (ICA) is the most commonly used tool in the neuroimaging community to investigate functional networks in the brain. With the advancement of imaging technology, neuroimaging studies with more complex study designs have become more commonly seen. Among them,longitudinal study has become a powerful way to investigate the changes in neural circuits with the progression of diseases or due to neurodevelopment. However, the existing group ICA methods cannot jointly decompose the repeatedly measured brain images from longitudinal fMRI studies.We proposed a longitudinal ICA (L-ICA) model that can appropriately incorporates the subject level random effects and also the time-dependent covariate effects in group ICA decomposition to capture the within-subject coherence and the time-evolving patterns in brain functional networks.
|
