Data-driven Cartesian sampling design for Compressed Sensing MRI
Frank Zijlstra 1 , Jaco J.M. Zwanenburg 1 , Max A. Viergever 1 , and Peter R. Seevinck 1
Image Sciences Institute, UMC Utrecht,
We propose a novel, data-driven method for optimizing
Cartesian undersampling patterns for Compressed Sensing.
The method iteratively adds sampling points based on CS
reconstructions of a training set. The performance of
the proposed optimized sampling patterns are evaluated
against the commonly used Variable Density undersampling
methods. Our method shows improvements in both the
Normalized Root Mean Square Error and the mean
Structural Similarity index. The method generalizes to
any reconstruction method that allows Cartesian
undersampling in any number of dimensions and would
enable optimization of patterns for a combination of CS
and parallel imaging.
This abstract and the presentation materials are available to members only;
a login is required.