Personalized Airway Trees from a Generative Model, Lung Atlas, and Hyperpolarized Helium MRI

Biomedical researchers are actively interested in building anatomically explicit computational models of the lung to further their understanding of pathologies, for example asthma, which affect the human body. Current methods have relied on a generic lung airway model which may not accurately reflect the physiology of a given subject. Patient specific models are needed to overcome this limitation. We propose a method for creating personalized models of the lung from Hyperpolarized Helium MRI images (Hyp. He MRI). As Hyp. He MRI images are of insufficient resolution to identify lung lobes, an estimate of lung lobe locations is obtained by registering a lung atlas to the MRI images. We then use an generative technique to create a lung airway model within the estimate lung lobe volumes. Initial testing indicates significant differences in predictions of lung function between the personalized model our approach generates and a generic airway model.