Parallelized Reconstruction of Spiral Fourier Velocity Encoding MRI Data in Multicore Processors

Rosana Ribeiro Lima¹, Joao L. A. Carvalho¹

Fourier velocity encoding (FVE) may be useful in the assessment of valvular disease and of carotid wall shear stress, as it eliminates partial volume effects that are an issue in phase-contrast imaging. The scan-time of 2DFT FVE is prohibitively long for clinical use, but the spiral FVE method is substantially faster. FVE can be significantly accelerated using compressed sensing. However, iterative reconstruction of spiral FVE is time-consuming, because of non-Cartesian sampling and multidimensionality. The sequential implementation of reconstruction algorithms underuses the total capacity of multicore processors. We demonstrate a reduction of reconstruction time of spiral FVE data through parallel programming.