Rapid pulmonary 129Xe ventilation MRI of discharged COVID-19 patients with zigzag sampling

Purpose To demonstrate the feasibility of zigzag sampling for 3D rapid hyperpolarized Xe-129 ventilation MRI in human. Methods Zigzag sampling in one direction was combined with gradient-recalled echo sequence (GRE-zigzag-Y) to acquire hyperpolarized Xe-129 ventilation images. Image quality was compared with a balanced SSFP (bSSFP) sequence with the same spatial resolution for 12 healthy volunteers (HVs). For another 8 HVs and 9 discharged coronavirus disease 2019 subjects, isotropic resolution Xe-129 ventilation images were acquired using zigzag sampling in two directions through GRE-zigzag-YZ. Xe-129 ventilation defect percent (VDP) was quantified for GRE-zigzag-YZ and bSSFP acquisitions. Relationships and agreement between these VDP measurements were evaluated using Pearson correlation coefficient (r) and Bland-Altman analysis. Results For 12 HVs, GRE-zigzag-Y and bSSFP required 2.2 s and 10.5 s, respectively, to acquire Xe-129 images with a spatial resolution of 3.96 x 3.96 x 10.5 mm(3). Structural similarity index, mean absolute error, and Dice similarity coefficient between the two sets of images and ventilated lung regions were 0.85 +/- 0.03, 0.0015 +/- 0.0001, and 0.91 +/- 0.02, respectively. For another 8 HVs and 9 coronavirus disease 2019 subjects, Xe-129 images with a nominal spatial resolution of 2.5 x 2.5 x 2.5 mm(3) were acquired within 5.5 s per subject using GRE-zigzag-YZ. VDP provided by GRE-zigzag-YZ was strongly correlated (R-2 = 0.93, p < 0.0001) with that generated by bSSFP with minimal biases (bias = -0.005%, 95% limit-of-agreement = [-0.414%, 0.424%]). Conclusion Zigzag sampling combined with GRE sequence provides a way for rapid Xe-129 ventilation imaging.