Magnetic resonance fingerprinting methods quantify multiple MR parameters by matching the measured signal to a precomputed 5-dimensional dictionary in which each parameter is a dimension. Generating high-precision maps requires dense grids in each dimension, which is prohibitively expensive in memory and computation time. We propose B-spline interpolation of the dictionary to reduce the dictionary size and to enable efficient nonlinear least-squares fitting by gradient-based optimization methods. The method is shown to substantially reduce fitting error.

This presentation is part of Minisymposium “MS47 - Splines in Imaging (3 parts)”
organized by: Carolina Beccari (Dept. Mathematics, University of Bologna) , Virginie Uhlmann (EPFL, Lausanne) , Michael Unser (EPFL, Lausanne) .

Authors:

Willem van Valenberg (Quantitative Imaging Group, Delft University of Technology, Biomedical Imaging Group Rotterdam, Erasmus MC)

Stefan Klein (Biomedical Imaging Group Rotterdam, Erasmus MC)

Frans Vos (Quantitative Imaging Group, Delft University of Technology, Radiology, Academic Medical Center, Amsterdam)

Lucas van Vliet (Quantitative Imaging Group, Delft University of Technology)

Dirk Poot (Biomedical Imaging Group Rotterdam, Erasmus MC)

Keywords:

inverse problems, magnetic resonance fingerprinting, nonlinear optimization, quantitative mri