The presence of gas-tissue interfaces at the alveolar surface has historically been viewed a substantial obstacle to pulmonary MR imaging. This is due to the >9 ppm susceptibility difference between diamagnetic tissue and paramagnetic O2 in the alveolar airspaces that generates a short T₂^* of 1 ms or less, depending on field strength (1,2). Despite these challenges, the magnetic susceptibility of lung tissue should be considered in the context of pulmonary function—namely gas exchange and blood oxygenation. That is, the susceptibility difference between alveolar air and adjacent pulmonary tissues is determined by the regional partial pressure of O₂, which is of central physiological importance. Similarly, deoxyhemoglobin (paramagnetic, with 4 unpaired Fe³⁺ electrons) becomes diamagnetic when oxygenated, a process that occurs efficiently only in regions where ventilation and pulmonary perfusion are properly matched. Thus, mapping the magnetic susceptibility of the lung tissue has the potential to provide fundamental insights into lung function in health and disease. A promising approach for overcoming the intrinsically short T₂^* of lung tissue, while simultaneously measuring its magnetic properties, is to perform quantitative susceptibility mapping (QSM) with short, multi-echo radial MRI (3). Here we demonstrate that: 1) when obtained with sufficiently short echo times, QSM is able to transform the typically negative T₂^* contrast observed in lungs into positive susceptibility signal and 2) the intensity of the positive susceptibility signal in the lungs is sensitive to O₂ partial pressure.

Animals: Free breathing C57BL/6 mice (n=6, ~25 g) were anesthetized with isoflurane mixed with medical grade air or pure O₂ gas. Respiratory motion was tracked with a pressure bellows placed on the abdomen, and body temperature was maintained at ~36°C with flowing warm air.

MRI: Images were acquired at end expiration using a 7T Bruker Biospec scanner and a custom-built quadrature birdcage coil. Imaging was performed with an RF-spoiled, 3D multi-echo radial sequence that employed interleaved, golden-angle trajectories. Imaging parameters include: radial views=51472; TR=9 ms; TEs=80, 200, 300, 400 and 500 μs; BW=278 kHz, FOV=40×40×60 mm3, matrix=128×128×128, resolution=313×313×469 μm³. For image reconstruction, k-space data were re-gridded using an iterative algorithm (4,5).

Quantitative susceptibility mapping: Phase was calculated from re-gridded k-space data. The tissue phase was calculated using Laplacian-based phase unwrapping and background phase removal algorithm called V-SHARP (6). The tissue phase was then inverted using the two-level QSM reconstruction algorithm, STAR-QSM (7). Finally, the quantitative susceptibility maps obtained from each of the echoes were summed to enhance susceptibility SNR.

Image analysis: The lung volume within the thoracic cavity was manually segmented from the body using Amira. Subsequent analyses were performed in MATLAB and ImageJ. Within the summed echo image, a region of interest (ROI) was measured in the skeletal muscle near the lungs to provide a reference susceptibility value. The susceptible value within this ROI was then subtracted from susceptibility values within the lung mask to generate background-subtracted images.

Relative to surrounding tissues, positive susceptibility signal was obtained from the lungs of all animals (e.g., see Fig. 1). When mice breathed air (~20% O2), susceptibility values across the lung volume varied by ~11 ppm and ranged from approximately -4 to 6 ppm. When animals breathed 100% O2 rather than air, the mean susceptibility increased from 1.6 to 2.0 ppm. Additionally, regional heterogeneity in susceptibility values, as assessed by the coefficient of variation (CV, standard deviation divided by mean), decreased by ~10% when the animal breathed pure O2.The positive susceptibility signal observed from lung tissue relative to extra-pulmonary tissues is consistent with the expected susceptibility shift caused by the presence of paramagnetic molecular O₂ gas. This interpretation of the data is further supported by the ~0.4 ppm increase in mean susceptibility observed from the lungs when mice breathed pure O₂ vs air. Note that no effort was made to exclude the pulmonary vasculature during image analysis. Thus, the simultaneous increase in susceptibility and decrease in regional heterogeneity (i.e., narrowing of the distribution of susceptibility values) observed when mice breathed pure O₂ likely reflects decreased deoxyhemoglobin content in the blood returning to the lungs via the pulmonary arteries.Quantitative susceptibility mapping of the lungs is feasible using MRI sequences with sufficiently short echo times. The positive contrast observed in lung QSM is sensitive to changes in pulmonary O₂ partial pressure and likely also to blood oxygenation. Together, these results demonstrate that multi-echo, radial QSM holds the potential to become a powerful probe of regional pulmonary function.