Pulmonary MRI morphometry modeling of airspace enlargement in chronic obstructive pulmonary disease and alpha‐1 antitrypsin deficiency
Diffusion‐weighted MRI apparent diffusion coefficients (ADCs) 4 reflect the Brownian motion of inhaled gas, and because these are relatively or completely inert (the small fraction of dissolved 129Xe may be ignored 5), gas displacement measurements reflect lung microstructural geometries. Unfortunately, ADC values cannot provide estimates of alveolar dimensions, and although these reflect alveolar size, more advanced diffusion‐weighted MRI approaches are required 1. A number of different acinar airway models, some of which are in silico, have been previously described 7, but for the cylindrical model 12 a very complete description based on empirical equations was derived 13 to assist in the quantitative analysis of the complex diffusion attenuations specific to gas diffusion in the lung microstructure 15. Using this cylindrical model of the acinar airways, a mathematical framework was previously generated 5, and as shown in Figure 1a, the transverse (DT) and longitudinal diffusion coefficients (DL) may be experimentally derived, reflecting the acinar duct external radius (R), internal radius (r), alveolar length (L), and alveolar sleeve depth (h) 16. Importantly, estimates may also be generated for the mean linear intercept (Lm) and surface‐to‐volume ratio (S/V) 16, both well‐described in the lung stereology literature 17. Previous work also established the relationship of lung MRI and histological measurements (16,18), and evaluated emphysema patients 16, rodent models 18, lung cancer 22, and lung aging 23. In contrast, a single‐compartment model, as shown in Figure 1b, was also previously used to explain fluid diffusion in porous media 25 and to model 129Xe 26, 19F 27, and simulated inhaled‐gas diffusion 28 in emphysematous tissue 29. Another approach to modeling gas displacement uses a stretched‐exponential model 30, in which MRI diffusion times and acinar geometries are not constrained. Until now, however, these diverse approaches have not been compared in the same patients, nor across a spectrum of parenchyma tissue enlargement/destruction.
Therefore, in this proof‐of‐concept demonstration, we evaluated three modeling approaches in elderly volunteers and patients with emphysema using single‐breath diffusion‐weighted MRI. We hypothesized that different modeling approaches would be required to estimate the different lung tissue abnormalities observed in senile, mild, and severe emphysema.