Pulmonary blood flow quantification in humans from 15O-water PET.

Abstract

Purpose: Dynamic positron emission tomography (PET) imaging has commonly been applied to study blood perfusion in the human brain and heart, but there is a very limited amount of existing research about the suitability of this method for many other organs of interest. Here, we focus on the quantification of pulmonary blood flow (PBF) in human lungs. We evaluate both the potential of the $_{}^{15}$ O-water PET imaging via compartmental modeling with automatic volume of interest (VOI) selection for PBF quantification and study the possible differences in PBF caused by different patient characteristics such as age or sex.

Procedures: We systematically fit the one-tissue compartment model to the mean time-activity curves derived from the $_{}^{15}$ O-water PET data of 103 patients. The machine learning-based segmentation tool TotalSegmentator is utilized to find segmentation masks for different lung lobes and right ventricle of the heart. Additionally, we automatically remove the majority of the air inside the lung lobe VOIs and the areas surrounding subclavian arteries and brachiocephalic veins with the help of binary erosion and dilatation operations. After the model fitting, we evaluate possible differences in the results caused by age, sex, weight, and body mass index (BMI) by performing Mann-Whitney U tests between different patient subgroups and computing Spearman's correlations coefficients.

Results: The estimated PBF within all the lung lobes had a mean of1.21±0.825 mL/min/cm $_{}^3$ and a median of 1.03 mL/min/cm $_{}^3$ , but this value was notably lower in right lower lung lobe and much higher in the upper lung lobes. The PBF was higher in both the female patients and in the patients under 65 years but not statistically significantly so. The individual variation was very high.

Conclusions: The PBF quantification based on $_{}^{15}$ O-water PET imaging combined with our automatic VOI selection method is an effective method to produce relatively realistic results. In case of upper lung lobes, the results are likely overestimated if pulmonary vessels are not removed from the VOI. The accurate estimation of the air volume within the lung lobe VOIs is also a non-trivial problem. More research on this topic is warranted to find whether there is a decreasing trend between PBF and age or significant differences between the sexes.