Joint estimation and correction of motion and geometric distortion in segmented arterial spin labeling

Purpose: Arterial spin labeling allows noninvasive measurement of cerebral blood flow by magnetically labeling inflowing blood, using it as endogenous tracer. Unfortunately, sensitivity to subject motion is high due to the subtractive nature of arterial spin labeling, which is especially problematic if Cartesian segmented 3D gradient and spin echo (GRASE) is applied. Using a 3D GRASE PROPELLER (3DGP) segmentation, retrospective correction of in-plane rigid body motion is possible before final combination of different segments. However, the standard 3DGP reconstruction is affected by off-resonance effects and has not yet been validated with different motion patterns and levels of background suppression.

Methods: The standard algorithm (1) and a Cartesian segmented 3D GRASE (2), as well as a new 3DGP reconstruction algorithm, which allows joint estimation of motion and geometric distortion (called 3DGP-JET), are validated in 5 healthy volunteers. Image quality of perfusion-weighted images was investigated for background suppression levels of 0%, 5%, and 10% in combination with no motion, as well as slow and fast intentional motion patterns during the scan.

Results: The proposed 3DGP-JET algorithm allowed robust estimation of field maps and motion for all scenarios, and greatly reduced motion-related artifacts in perfusion-weighted images when compared with Cartesian segmented 3D GRASE.

Conclusion: Further improvements of the presented 3DGP-JET routine and a combination with prospective motion correction are recommended to compensate for through-plane motion, making the presented technique a good candidate for dealing with motion-related artifacts in arterial spin labeling images in clinical reality.