gpuNUFFT - An Open-Source GPU Library for 3D Gridding with Direct Matlab Interface

A fast 3D NUFFT CUDA implementation with Matlab mex interface and class based generation of forward and adjoint operators. It can be integrated in any Matlab based image reconstruction and will perform the NUFFTs on the GPU without having to change anything else in the code.

Contributors:

• Andreas Schwarzl (Graz University of Technology)
• Florian Knoll

Version History:

May 2016: The download link from this page is removed now. The most recent version of the library is now available directly from GitHub. gpuNUFFT is now also featured on the ISMRM MR-Hub. Additional installation instructions, video tutorials and an ISMRM review of the library can be accessed from there.

Note (August 2015): Compute capabilities 1.x are no longer supported in CUDA 7.0. When using gpuNUFFT with CUDA 7.0, the CMakeLists file has to be adjusted. The flags SET(CUDA_NVCC_FLAGS "-arch;sm_1x;--ptxas-options=-v") have to be modified. We will adress this more thoroughly in future releases.

2.0.6 (September 15th 2014): Major update. All precomputations done on the GPU, full support for multi coil data, support for arbitrary sector widths, anisotropic grid sizes supported effectively.

1.1 (May 28th 2014): Changed demo example to allow use of atomic operations for increased speedup.

1.0 (May 5th 2014): First release of gpuNUFFT.

Related Publication:

If you use this library or the provided data in your publications, please cite:

Knoll, F.; Schwarzl, A,; Diwoky, C.; Sodickson DK.: gpuNUFFT - An Open-Source GPU Library for 3D Gridding with Direct Matlab Interface. Proc ISMRM p4297 (2014).