To establish thresholds for healthy lung volumes on CT, we analyzed a subcohort from an ongoing ultra-low-dose chest photon-counting CT study focused on incidental lung nodules during coronary CT angiography (n=659). 

We employed rigorous criteria to define "healthy subjects," which included a BMI of less than 35, an Agatston Calcium Score of less than 100, and no history of smoking, asthma, or dyspnea. After manual revision of the remaining 131 subjects, we excluded individuals with signs of prior chest surgery (either cardiac or pulmonary), significant chest deformities, severe hyperkyphosis or scoliosis, phrenic paresis, or a wide mediastinum due to any reason (such as lung abnormalities, solid or cystic tumors). We also excluded scans that were inadvertently captured during expiration and those exhibiting hyperinflated lungs. 

Lung volumes were quantified using deep-learning-based CAD software (contextflow, Vienna, Austria), which we also used to segment visually unremarkable pathologies automatically. Additionally, we excluded patients with detected emphysema of more than 1% of measured lung volume. We also applied a cumulative exclusion threshold for reticulations, consolidation, and emphysema of more than 3%.

Finally, we developed a nomogram using a GAMLSS model validated in lung function testing. To assess normality, we generated worm plots in RStudio.