Removal of off-resonance xenon gas artifacts in pulmonary gas-transfer MRI.

PURPOSE: Hyperpolarized xenon (129 Xe) gas-transfer imaging allows different components of pulmonary gas transfer-alveolar air space, lung interstitium/blood plasma (barrier), and red blood cells (RBCs)-to be assessed separately in a single breath. However, quantitative analysis is challenging because dissolved-phase 129 Xe images are often contaminated by off-resonant gas-phase signal generated via imperfectly selective excitation. Although previous methods required additional data for gas-phase removal, the method reported here requires no/minimal sequence modifications/data acquisitions, allowing many previously acquired images to be corrected retroactively. METHODS: 129 Xe imaging was implemented at 3.0T via an interleaved three-dimensional radial acquisition of the gaseous and dissolved phases (using one-point Dixon reconstruction for the dissolved phase) in 46 human subjects and a phantom. Gas-phase contamination (9.5% ± 4.8%) was removed from gas-transfer data using a modified gas-phase image. The signal-to-noise ratio (SNR) and signal distributions were compared before and after contamination removal. Additionally, theoretical gaseous contaminations were simulated at different magnetic field strengths for comparison. RESULTS: Gas-phase contamination at 3.0T was more diffuse and located predominantly outside the lungs, relative to simulated 1.5T contamination caused by the larger frequency offset. Phantom experiments illustrated a 91% removal efficiency. In human subjects, contamination removal produced significant changes in dissolved signal SNR (+7.8%), mean (-1.4%), and standard deviation (-2.3%) despite low contamination. Repeat measurements showed reduced variance (dissolved mean, -1.0%; standard deviation, -8.4%). CONCLUSION: Off-resonance gas-phase contamination can be removed robustly with no/minimal sequence modifications. Contamination removal permits more accurate quantification, reduces radiofrequency stringency requirements, and increases data consistency, providing improved sensitivity needed for multicenter trials.