Comparison of gradient delay measurement and correction in diffusion-weighted 3D cones imaging in the breast

Motivation: Gradient delay correction must be carefully considered in non-cartesian imaging such as 3D cones to avoid severe artifacts. Goal(s): To compare in-vivo-based with phantom-based delay measurement and correction in diffusion-weighted 3D cones breast MRI. Approach: In 88 subjects, two blinded readers quality-ranked reconstructed images of double-echo steady-state diffusion-weighted breast MRI with three-dimensional cones using (i) no correction, (ii) in-vivo-based delay correction, and (iii) phantom-based delay correction. Results: Reconstructions with in-vivo-based correction were ranked best in only 40% of cases. Reconstructions done with phantom-based correction were ranked best in 89% of cases. Generally, in 94% of cases, correction led to improved image quality. Impact: In non-cartesian readouts, gradient delays cause artifacts. For diffusion-weighted breast MRI with a 3D cones trajectory, we show that retrospective delay correction based on in-vivo measurements requires further development to be considered a robust method.