The University of Iowa           CS Home > Events > Colloquia

Search UI CS Site

[an error occurred while processing this directive]

Spotlight on HPC Applications in Medical Imaging Jun Ni Academic Technologies-Research Services Information Technology Services The University of Iowa Friday, October 20, 2006 4:00-4:50pm, 61 SH Abstract This presentation demonstrates how the ITS Scientific Computing Group (SCG) supports University of Iowa medical imaging research using high-performance computing. The presenter begins by outlining the mission and responsibilities of the SCG group. Then two representative projects are discussed. These multi-disciplinary projects have attracted significant external funds to the University. The first is the NIH-funded Digital Lung Project for multi-scale simulations of gas flow in human lungs. This project required development of parallel algorithms of Lattice Boltzman Methods (LBM) in modeling and simulating bubble airflow. The parallel LBM algorithm was designed and implemented on various HPC systems and/or in a distributed computing environment. The experience gained from this project provides a concrete foundation for designing and conducting large-scale, intensive computations using TeraGrid resources. The deployment of parallel computing technology accelerated the project, enhancing our prospects for future NIH grant applications. The second project, also funded by NIH, is the systematic development of parallel algorithms of large-scale CT medical image reconstruction. Although algorithms for generating CT patient images have improved over the years, one challenge remains: finding an economical and efficient way to quickly perform medical image reconstruction, especially for high-resolution datasets or micro CT image processing. The goal is to use today's parallel computing and Grid computing technologies to speed up future image reconstruction for radiological diagnostics. Following a description of today's parallel CT image reconstruction algorithms is an introduction of the Katsevich algorithm and its parallel implementation on various HPC systems, with emphasis on the promising results achieved to date. Future objectives are also highlighted. These two projects provide great opportunities for exploring unprecedented cyberinfrastructure-based medical imaging systems and networks.

[an error occurred while processing this directive].

	Translate this page automatically.
	©2005 The University of Iowa, All Rights Reserved.