Intrinsic correction of system delays for radial magnetic resonance imaging

When using radial MR image acquisition techniques gradient or sampling delays due to hardware imperfections can cause mismatch between the expected and the actual k-space trajectory along the readout direction. To provide a robust and simple correction of such system delays we developed a new calibration method which is independent of using any reference data or applying sequence modifications.

Radial data obtained with 180°, 360° and golden-angle radial ordering schemes were deliberately shifted along the readout direction for a discrete range of gradient delays. Following 2D regridding, images were reconstructed and analyzed in image space for all applied shifts to estimate the optimal system delay. Phantom and in vivo measurements were performed to test the robustness of the algorithm.

Using the 360° and golden-angle radial ordering schemes system delays in the range of 3.3μs to 6.3μs were estimated and corrected for several imaging applications and different conditions, including cardiac and real-time MRI as well as multiple acquisitions using different imaging parameters and slice orientations. When using the standard 180° radial acquisition scheme no automated correction was possible. With a mean computation time of 23.2±14.0s for the delay estimation computational demands were moderate allowing implementation of the algorithm on the image reconstruction system of any modern MR system.

We have demonstrated that radial data acquired with a 360° or golden-angle ordering scheme can be used for reliable intrinsic correction of system delays. The proposed technique enables a per-scan correction of system delays without the need for additional calibration data or modifications of the radial imaging sequence.