Abstract Rationale Conventional methods of assessing 129Xe ventilation use histogram-based techniques such as rescaling to arbitrary thresholds or image-based clustering to estimate the ventilation defect percent (VDP). However, histogram rescaling is highly sensitive to the range of voxel intensities and yields limited interpretability of individual voxel values. To that end, we previously demonstrated an alternative approach to derive single-breath fractional ventilation (FV) maps. This approach directly estimates voxel-specific FV based on known inhaled xenon gas volume, thereby addressing both rescaling and providing a physiologically interpretable measure of regional ventilation. Here, we extend this method to ventilation measured via gas exchange 129Xe MRI and evaluate its same-session repeatability. We use this method to calculate VDP (VDP-FV) and compare it to that derived by conventional 99th percentile histogram rescaling (VDP-99). Methods 27 participants with idiopathic pulmonary fibrosis (IPF), (9F, age 72.4±6.5), and 22 healthy participants (9F, age 41.1±14.9) underwent repeated same-session scans 129Xe gas exchange imaging with N4ITK bias correction. FV was calculated for each participant by dividing the total xenon signal measured within the field of view by the known volume of the inhaled xenon dose. The resulting “xenon signal to volume” ratio was then used to estimate voxel-wise FV, defined as the volume of inhaled xenon mixture per voxel divided by the voxel volume. Repeatability of the mean FV metric was evaluated in the combined healthy and IPF cohort using Bland-Altman analysis, the intraclass correlation coefficient (ICC), and the coefficient of repeatability (CR). VDP-FV was computed as the percentage of the histogram ≥2sd below a young healthy reference mean, and similarly compared with VDP-99. Results Mean FV in the healthy participants was 0.229±0.0467 vs. 0.182±0.0342 in IPF (p 0.001). For mean FV, ICC=0.876 and CR = 0.043. For VDP-FV, ICC=0.912 and CR = 10.4%, vs. VDP-99 with ICC=0.881 and CR = 6.5%. The limits of agreement (LoA) for VDP-FV vs. VDP-99 with combined same-session scans were -15.3%-21.5%, bias = 3.12, and ICC=0.56. Conclusions Our proposed FV method provides physiologically plausible estimates of regional fraction ventilation based on first principles. It demonstrates strong within-session repeatability, comparable to that of conventional 99th percentile histogram rescaling, while also enabling direct interpretation of voxel values as the fraction of gas turned over locally. Its consistency and interpretability make it a promising method for assessing ventilation heterogeneity. Ongoing work will further improve the robustness and repeatability of this measurement and characterize FV patterns in other populations. This abstract is funded by: NIH
Du et al. (Fri,) studied this question.