Human airway material characterization via inverse finite element analysis and neural network surrogate.

Central airway obstruction can be caused by lung cancer and may severely diminish respiratory function to necessitate airway stenting. However, the mechanical properties of airway tissues remain poorly characterized, leading to a mismatch between stent mechanical behavior and airway compliance, which can reduce stent biocompatibility and clinical effectiveness. Such mismatches can result in abnormal stress transfer at the stent-airway interface and cause stent migration, local tissue irritation or inflammation, and impaired long-term performance. Predictive models are often used in treatment planning, however, without a comprehensive understanding of airway properties, advancements in modeling remain highly limited. In this study, we develop an experimental to numerical pipeline for identifying the mechanical properties of the human airway branching network, specifically the trachea, right bronchus, and left bronchus tissues. Biaxial planar tensile tests are used to capture the tissues' unique, anisotropic mechanical response in order to inform the computational routine-where inverse finite element analysis is employed to calibrate a Holzapfel-Gasser-Ogden constitutive model. To overcome the computational burden of traditional IFEA, a neural network (NN) surrogate model is trained to enable rapid material identification. We identify parameter values for the various tissues, such as C