Evaluation of functional connectivity in control and ADHD subjects by generalized Canonical Correlation and Probabilistic Independent Component Analysis

Functional connectivity can be evaluated by temporal correlation between spatial neurophysiologic events or correlation between neural activities of brain regions. Unlike anatomical connectivity which represents physical connections between two brain regions, functional connectivity investigates interactions between brain regions. In this study, we implemented and evaluated Probabilistic Independent Component Analysis (PICA) and generalized Canonical Correlation Analysis (gCCA) on resting state data of healthy subjects and subjects with ADHD disorders from NITRC database. This was done to detect brain functional connectivity and compare the results in normal and diseased subjects. In addition, speed of the methods was compared. Our study shows that relative to PICA, gCCA has higher precision and accuracy and lower computational complexity and run time in addition to capability of extracting more reproducible statistical maps in both control and ADHD subjects.