3D magnetic resonance imaging of the human brain-novel radial sampling, filtering and reconstruction

Abstract

We have suggested a novel method PRESTO-CAN including radial sampling, filtering and reconstruction of k-space data for 3D-plus-time resolved MRI. The angular increment of the profiles was based on the golden ratio, but the number of angular positions N was locked to be a prime number which guaranteed fix angle positions. The time resolution increased dramatically when the profiles were partly removed from the k-space using the hourglass filter. We aim for utilizing the MRI-data for fMRI, where the echo times are long, TE ≈ 37-40 ms. This will result in field inhomogeneities and phase variations in the reconstructed images. Therefore, a new calibration and correction procedure was developed. We show that we are able to reconstruct images of the human brain with an image quality in line with what can be obtained by conventional Cartesian sampling. The pulse sequence for PRESTO-CAN was implemented by modifying an existing PRESTO sequence for Cartesian sampling. The effort involved was relatively small and a great advantage will be that we are able to use standard procedures for speeding up data acquisition, i.e. parallel imaging with SENSE.