The July 2010 Meeting of the San Francisco Bay Area Chapter

         of the American Statistical Association (ASA)

 

Date and Time:           Tuesday, July 27th, 2010  4:00 - 5:30 PM

Agenda:                          (a) talk by Dan Rowe

Speaker:                         Dan Rowe, Marquette University

Title:                                A Complex-Valued Mathematical Model for fMRI Analysis

 

Light refreshments will be served 4:00 - 4:30 PM. The talk will begin at 4:30 PM. The speaker will be taken to dinner afterwards. Everyone is welcome to join.

 

Location:                        Center for Molecular and Functional Imaging (CMFI) classroom, Room #331,  China Basin Landing (CBL),  Lobby 6, 3rd Floor,  Department of Radiology, University of California, San Francisco (UCSF) 185 Berry Street, San Francisco. Driving Directions and Map. Lobby 6 and Classroom #331 are located near the western end of CBL, across from the San Francisco Public Library on 4th Street.

 

Abstract:                        In this talk I will present a single coherent mathematical framework for functional magnetic resonance imaging (fMRI) Analysis. In fMRI, it is necessary to apply many preprocessing and transformation operations to the originally measured complex-valued measurements. The physics signal equation describes the original raw unprocessed complex-valued measurements. The standard preprocessing and transformation operations are not incorporated into the modeling and analysis. As a result, there is a complete disconnect between the analysis models and the underlying physics of the biological system being imaged. The framework that I have developed incorporates all of the standard preprocessing and transformations of fMRI measurements within a single coherent mathematical model. Starting with the physics signal equation, the standard preprocessing and transformation operations on the raw complex-valued data will be described using a matrix algebraic framework. The complex-valued fMRI activation models that I have developed are generalized to include spatio-temporally correlated time series measurements as would result from preprocessing. Results from complex-valued brain activation models will be presented and related to the fundamental physical quantities of proton spin density, transverse relaxation, and magnetic field inhomogeneities.

 

 Return to Bay Area ASA Homepage