Improved fat–water reconstruction algorithm with graphics hardware acceleration

Abstract

Purpose: To develop a fast and robust Iterative Decomposition of water and fat with Echo Asymmetry and Leastsquares (IDEAL) reconstruction algorithm using graphics processor unit (GPU) computation. Materials and Methods:The fat-water reconstruction was expedited by vectorizing the fat-water parameter estimation, which was implemented on a graphics card to evaluate potential speed increases due to data-parallelization. In addition, we vectorized and compared Brent's method with golden section search for the optimization of the unknown field inhomogeneity parameter (c) in the IDEAL equations. The algorithm was made more robust to fat-water ambiguities using a modified planar extrapolation (MPE) of c algorithm. As compared to simple planar extrapolation (PE), the use of an averaging filter in MPE made the reconstruction more robust to neighborhoods poorly fit by a two-dimensional plane.Results: Fat-water reconstruction time was reduced by up to a factor of 11.6 on a GPU as compared to CPU-only reconstruction. The MPE algorithms incorrectly assigned fewer pixels than PE using careful manual correction as a gold standard (0.7% versus 4.5%; P < 10 À4 ). Brent's method used fewer iterations than golden section search in the vast majority of pixels (6.8 6 1.5 versus 9.6 6 1.6 iterations).Conclusion: Data sets acquired on a high field scanner can be quickly and robustly reconstructed using our algorithm. A GPU implementation results in significant time savings, which will become increasingly important with the trend toward high resolution mouse and human imaging.

Hepatitis C trials that combine investigational agents with pegylated interferon should be stratified by interleukin-28B genotype

“…Many attempts have been made to resolve the ambiguity using the a priori knowledge of B0 field map (ψ) smoothness, based on the observation that real magnetic fields do not have discontinuities. Region‐growing schemes (11–14) have been used to generate initial guesses at each pixel. However, these algorithms are prone to error propagation, because the initialization at a pixel depends on the previously solved pixels.…”

Body composition analysis of obesity and hepatic steatosis in mice by relaxation compensated fat fraction (RCFF) MRI

“…The multiecho aRARE acquisition also reduced the overall acquisition time in comparison to conventional asymmetric spin echo acquisitions to provide three coregistered image sets with an acquisition time of ≈15 minutes. The modified 7T IDEAL reconstruction used in this study used Brent's method and planar extrapolation to accurately delineate fat and water components of 20–25 imaging slices in ≈5 minutes (19, 23). RCFF maps were then calculated from the water and fat magnitude images by Eq.…”

“…The aRARE sequence provided a user‐selectable echo delay to achieve π/6, 5π/6, and 3π/2 radian shifts between fat and water (echo shifts of 79, 396, and 714 μs at 7T). The MRI images obtained from the aRARE acquisition described above were used to generate separate fat and water images by an modified Iterative Decomposition of Water and Fat with Echo Asymmetry and Least‐Squares Estimation (IDEAL) image reconstruction technique for 7T described previously (18, 19).…”

Cryo-image Analysis of Tumor Cell Migration, Invasion, and Dispersal in a Mouse Xenograft Model of Human Glioblastoma Multiforme

“…Processing times of Matlab codes can be reduced by converting to C++. Alternatively, many of our algorithms use image filtering, which is easily accelerated using GPU processing as has recently been done by us for another application [19]. …”