Correction of Respiratory Motion in Free-Breathing DCE-MRI Using a Pilot-Tone Coil

Respiratory motion poses a major challenge in abdominopelvic MRI and often leads to strong artifacts. Self-navigation techniques have been proposed for motion correction, allowing free-breathing examination. However, self-navigation methods typically fail to track respiratory motion for dynamic contrast-enhanced (DCE) scans due to concurrent signal changes caused by the contrast injection. In this study, we investigated the use of a pilot-tone (PT) transmitter integrated into the receive coil as an alternative approach for obtaining respiratory curves. The amplitude of the received PT signal gets modulated by respiratory motion due to changes in the coil load. A pipeline for extracting respiratory curves from the PT signal was developed and applied for motion-corrected free-breathing abdominal DCE-MRI. Validation experiments included real-time cine imaging and comparison with self-navigation for five late-phase liver scans. Dynamic experiments were performed for a liver DCE scan of one healthy volunteer and kidney DCE scans of 24 patients. A motion-weighted GRASP algorithm was used for reconstructing dynamic images. Quantitative metrics, including gradient entropy for individual images and the L2 norm of the temporal differences for each dynamic frame set, were calculated. Motion-weighted DCE GRASP reconstructions using PT-derived respiratory curves showed improved image sharpness, especially in regions such as the diaphragm, kidney boundaries, and vessels, as well as better temporal consistency compared to conventional GRASP reconstructions. These improvements were observed both qualitatively and quantitatively. This work demonstrates that the PT technique can accurately detect respiration during contrast injections, reducing motion blurring in free-breathing dynamic DCE-MRI. The proposed technique can be easily implemented in the clinical workflow by using a coil with an integrated PT transmitter and requires no external belts or sensors.