Prospective correction of patient‐specific respiratory motion in myocardial T 1 and T 2 mapping

Purpose Respiratory motion in cardiovascular MRI presents a challenging problem with many potential solutions. Current approaches require breath‐holds, apply retrospective image registration, or significantly increase scan time by respiratory gating. Myocardial T 1 and T 2 mapping techniques are particularly sensitive to motion as they require multiple source images to be accurately aligned prior to the estimation of tissue relaxation. We propose a patient‐specific prospective motion correction (PROCO) strategy that corrects respiratory motion on the fly with the goal of reducing the spatial variation of myocardial parametric mapping techniques. Methods A rapid, patient‐specific training scan was performed to characterize respiration‐induced motion of the heart relative to a diaphragmatic navigator, and a parametric mapping pulse sequence utilized the resulting motion model to prospectively update the scan plane in real‐time. Midventricular short‐axis T 1 and T 2 maps were acquired under breath‐hold or free‐breathing conditions with and without PROCO in 7 healthy volunteers and 3 patients. T 1 and T 2 were measured in 6 segments and compared to reference standard breath‐hold measurements using Bland‐Altman analysis. Results PROCO significantly reduced the spatial variation of parametric maps acquired during free‐breathing, producing limits of agreement of −47.16 to 30.98 ms (T 1 ) and −1.35 to 4.02 ms (T 2 ), compared to −67.77 to 74.34 ms (T 1 ) and −2.21 to 5.62 ms (T 2 ) for free‐breathing acquisition without PROCO. Conclusion Patient‐specific respiratory PROCO method significantly reduced the spatial variation of myocardial T 1 and T 2 mapping, while allowing for 100% efficient free‐breathing acquisitions.