Greg Baxter


Senior Programmer
B.S Mechanical Engineering, University of Illinois, Urbana-Champaign
Ph.D. Mechanical Engineering, University of California, Berkeley

Research Interests
My main research interest is modeling diffusion in realistic biological tissues. The goal of our research in this area is to understand the processes underlying diffusion tensor imaging (DTI) and to be able to improve image acquisition and analysis. Clinical DTI scans are limited to relatively short times (30 minutes). With current hardware, this time limitation forces trade-offs in spatial resolution (the ability to differentiate nearby structures), angular resolution (the ability to differentiate motion along similar directions), or the volume imaged. By better understanding the diffusion and imaging processes, we hope to be able to produce imaging and analysis methods with better diagnostic capabilities.

Research Projects
My main research project is DifSim, our diffusion simulator toolkit, which is designed to allow researchers to simulate an entire DTI experiment computationally. DifSim includes tissue generation tools for creating aligned fiber bundles (similar to white matter in the human brain) as well as surrounding ellipsoidal cells. Users can specify the pulse sequence parameters for a DTI simulation. DifSim simulates the diffusion of particles in the synthetic tissue by a physically accurate Monte Carlo random walk algorithm and calculates the resulting DTI signal. I am also involved in the development of the AFNI Diffusion Plugin, an extension for AFNI which provides analysis tools specific to diffusion tensor imaging, including calculation of derived quantities (Fractional Anisotropy, Mean Diffusion, etc.) and alternate representations (e.g. spherical harmonic decompositions or orientation distribution functions). In addition, the AFNI Diffusion Plugin provides a 3D viewer for visualizing diffusion directions (primary eigenvectors of the diffusion tensor) and fiber tractography.

Gregory T. Balls and Lawrence R. Frank. A Realistic DTI Simulation Environment. In Proc. ISMRM 16th Scientific Meeting, Toronto, Canada, May 2008. International Society of Magnetic Resonance in Medicine.

Peter McCorquodale, Phillip Colella, Gregory T. Balls, and Scott B. Baden. A Local Corrections Algorithm for Solving Poisson's Equation in Three Dimensions. Communications in Applied Mathematics and Computational Science Vol.2, No.1 (2007), pp 57-81.

Gregory T. Balls, Scott B. Baden, Tilman Kispersky, Thomas M. Bartol, and Terrence J. Sejnowski. A large scale Monte Carlo simulator for cellular microphysiology. Conf. Proc. IPDPS-2004, Santa Fe, NM, April 2004.

Gregory T. Balls, Scott B. Baden, Phillip Colella, SCALLOP: A Highly Scalable Parallel Poisson Solver in Three Dimensions. Conf. Proc. Supercomputing 2003, Phoenix, AZ, November 2003.