Medical Physics Seminar - Monday, April 28, 2008

QUICK LINKS: [Medical Physics Home Page] [Seminar Home Page]

A Comparison of Computational Methods to Calculate Effective Connectivity from fMRI Time Series Data

Suzanne Witt

Research Assistant

Department of Medical Physics

UW -School of Medicine and Public Health

Madison, Wisconsin

Functional magnetic resonance imaging (fMRI) has provided important insights to the study of neuroscience.  However, limitations in temporal resolution has led to the use of fMRI to address mainly where information is processed in the brain.  Additional information is needed to study how the brain processes information.  Many approaches have been investigated to study the functional integration of various cognitive and visuo-motor tasks, including effective connectivity. Effective connectivity, the influence of one neuronal system on another, attempts to address directionality of influence by defining an explicit statistical model of directed neuronal interactions.  Several different methods have been presented in the literature as a means to calculate the path weights of effective connectivity models.  As effective connectivity analyses become more common, it is important to understand how these computational methods compare with each other in terms of their abilities to detect changes in path weight values between tasks or subjects.  A comparison of three of the more popular methods – structural equation modeling, autoregressive analysis, and Granger causality will be presented. 

Location: 1335 Health Sciences Learning Center (HSLC)

Time:  4:00pm-5:00pm

Refreshments will be provided prior to the talk

QUICK LINKS: [Medical Physics Home Page] [Seminar Home Page]