Biostatistics seminar: Applications of Independent Component Analysis for MRI Data- 11/13/2018

Lisa Nickerson, PhD, McLean Hospital, will present several applications of independent component analysis (ICA) techniques, including the application of single-subject ICA for denoising fMRI data and group ICA for investigating functional connectivity of brain networks, highlighting the work on dual regression and findings on brain network connectivity in individuals with alcohol use disorder. Also, a new analytic method will be discussed that uses linked ICA to denoise scanner effects from multi-study MRI data, and the findings using linked ICA for multi-modal data fusion to investigate brain structure and function in heavy chronic cannabis users. Registration not required.

Lisa Nickerson, PhD; Assistant Professor of Psychiatry, HMS; Director of Applied Neuroimaging Statistics Laboratory, McLean Hospital

Abstract

Independent component analysis (ICA) is a purely data-driven multivariate statistical method that has found numerous applications in MRI data analysis. The ability of ICA to separate meaningful signals of interest from confounding noise signals in a data-driven fashion make it a powerful dual purpose technique for both addressing noise problems in MRI data and for investigating brain function. Dr. Nickerson will present several such applications of ICA techniques, including the application of single-subject ICA for denoising fMRI data and group ICA for investigating functional connectivity of brain networks, highlighting the work on dual regression and findings on brain network connectivity in individuals with alcohol use disorder. Also, a new analytic method will be discussed that uses linked ICA to denoise scanner effects from multi-study MRI data, and the findings using linked ICA for multi-modal data fusion to investigate brain structure and function in heavy chronic cannabis users.