Online Program Home
My Program

Abstract Details

Activity Number: 28 - SPEED: A Mixture of Topics in Health, Computing, and Imaging
Type: Contributed
Date/Time: Sunday, July 29, 2018 : 2:00 PM to 3:50 PM
Sponsor: Section on Statistics in Imaging
Abstract #330743
Title: Fusion of the Semiparametric Models and Network Measures in the Study of Brain Dynamic Functional Connectivity
Author(s): Maria Kudela* and Jaroslaw Harezlak and Mario Dzemidzic and Brandon Oberlin and David A Kareken and Joaquin Goni
Companies: Takeda Pharmaceuticals and Indiana University Bloomington and Indiana University School of Medicine and Indiana University School of Medicine and Indiana University School of Medicine and Purdue University
Keywords: dynamic functional connectivity; modularity analysis; graph theory; semiparametric models; networks
Abstract:

Functional MRI data can inform us about dynamically changing associations between distinct brain regions i.e., dynamic functional connectivity; dFC). A novel combination of the semiparametric models and network measures allowed us to quantify dFC changes during fMRI task.

Specifically, we utilized subject-specific nonparametric estimates of dFC (Kudela et al. 2017) in the additive mixed model framework to obtain the group level dFC estimates. Subsequently, we incorporated dFC estimates in the extension of traditional modularity analysis (Mucha et al. 2010) to get the dynamic division of brain regions into communities with intra-connectivity greater than expected by chance. This dynamic modularity was summarized on the population and subject level at a brain network and region basis by using entropy (a measure of uncertainty in module assignment during fMRI scan).

The above-described approach offers a low-dimensional representation that supports the existence of common functionally-based organization of the brain and provides biologically meaningful results. This novel methodology can be applied to both task and resting state fMRI data.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program