Phase sensitive imaging of dissolved-phase hyperpolarized ¹²⁹Xe holds great promise for directly and non-invasively imaging gas exchange impairment in idiopathic pulmonary fibrosis (IPF)^1,2. However, the limited and non-renewable hyperpolarization, low solubility of xenon in tissues and blood, its broad and overlapping resonances, and the rapid T2* decay pose major challenges to separating ¹²⁹Xe uptake in blood and barrier tissues in vivo. Here we build upon previous work using Dixon-based decomposition¹ by optimizing our reconstruction for the specific application of gas-exchange imaging³. We present examples that demonstrate the utility of such improved image quality for enabling quantitative analysis of gas exchange impairment in IPF. Specifically, this has enabled us to demonstrate that the ratio of ¹²⁹Xe uptake in the RBCs and barrier tissues differs significantly between the apex and base of the lung. Because fibrosis develops initially in the base of the lungs, the ability to detect functional changes in these regions may hold promise that ¹²⁹Xe MRI detects the earliest manifestations of disease.6 healthy controls and 7 IPF subjects underwent hyperpolarized ¹²⁹Xe imaging on a 1.5T GE 15M4 EXCITE MRI scanner (GE Healthcare, Waukesha WI) and a Quadrature ¹²⁹Xe vest coil (Clinical MR Solutions, Brookfield, WI). After inhaling an 85 mL dose equivalent⁴ of hyperpolarized ¹²⁹Xe (Model 9800, Polarean, Inc., Durham, NC), subjects held their breath for 15 seconds during the interleaved 3D radial acquisition that alternated between imaging the gas- and dissolved-phase resonances with scan parameters: FOV = 40 cm, TE/TR = ~0.9/7.5 ms, BW = 15 kHz, 64 samples/ray, 2002 rays, 1.2 ms 3-lobed sinc pulse, 0.5/22° flip angle (gas/dissolved). Gas- and dissolved-phase images were separately reconstructed at isotropic resolution of 6 mm. The reconstruction kernel sharpness was tuned (σ=0.14 for dissolved phase images and σ=0.32 for ventilation images) to maximize resolution and SNR despite the high degree of undersampling (1.94% of Nyquist requirement) and limited dissolved phase signal³. Next, the dissolved-phase image was decomposed into RBC and barrier images using previously described methods¹. From these image volumes, the RBC:barrier images were derived to quantify regional gas exchange in the apical vs. basal lung.Figure 1 demonstrates the improvement in spatial resolution that is afforded by properly tuning the reconstruction kernel for the given degree of undersampling and SNR of the measured data. Figure 2 illustrates two key features that became noticeable with the image quality improvements. First these improved images now readily reveal the gravitational gradient in gas exchange in healthy subjects. Second, the resolution is now sufficient to resolve the fissures between lung lobes in coronal slices. These improvements enable quantification of differences in the RBC:barrier ratio between the apex and base of the lung. Figure 3 shows that among IPF subjects, the RBC:Barrier reduced dramatically from apex to base (32 ± 19%) compared to only 15 ± 14% seen in healthy subjects (p=0.05).This work demonstrates technical improvements to Dixon-based decomposition of dissolved-phase ¹²⁹Xe imaging that enable differentiation between IPF subjects and healthy controls based on RBC:barrier ratios in the basal and apical lung. This reduction of gas exchange in the basal lung is consistent with the known distribution of fibrosis in IPF, and is thus a useful step in validating the accuracy of Dixon-based decomposition. Because the gas exchange impairment associated with IPF is typically heterogeneous, imaging approaches such as ¹²⁹Xe MRI are particularly well suited for detecting and monitoring disease progression and therapy response.