Abstract #0525

PowerGrid: A open source library for accelerated iterative magnetic resonance image reconstruction

Alex Cerjanic^1,2, Joseph L Holtrop^1,2, Giang Chau Ngo¹, Brent Leback³, Galen Arnold⁴, Mark Van Moer⁴, Genevieve LaBelle^2,5, Jeffrey A Fessler⁶, and Bradley P Sutton^1,2

¹Bioengineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ²Beckman Institute, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ³PGI Compilers & tools; an NVIDIA brand, Portland, OR, United States, ⁴National Center for Supercomputing Applications, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ⁵Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ⁶Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI, United States

PowerGrid is an accelerated, open source, freely available toolkit for iterative reconstruction supporting non-Cartesian trajectories. Using high level compiler directives, GPU accelerated Fourier transform operators were implemented in a high level syntax designed to correlate with the popular Image Reconstruction Toolbox (IRT). A speed-up of up to 8.96x over the unaccelerated IRT reconstruction was obtained using an NVIDIA Tesla K40c accelerator.

This abstract and the presentation materials are available to members only; a login is required.

Join Here