Uncertainty aware virtual treatment planning for peripheral pulmonary artery stenosis

点击翻译按钮获取中文摘要

Congenital heart disease (CHD) affects 1/100 babies and is the leading cause of infant mortality in the U.S. Pulmonary artery (PA) stenosis is common in CHD patients and is particularly challenging to treat when occurring in the periphery of the PA tree. Peripheral pulmonary stenosis (PPS), often consisting of numerous vessel narrowings at proximal and distal bifurcation levels, can lead to persistent RV hypertension, RV failure, and even death. Most institutions treat PPS patients with stenting and angioplasty limited to the proximal (central and lobar) PAs only. These catheter-based interventions, however, are often ineffective at reducing right ventricular pressures and are associated with poor and unpredictable outcomes. Comprehensive surgical reconstruction, involving patch augmentation of ALL stenoses (central, lobar, segmental PAs), can achieve long-term RV pressure reduction with low morbidity and mortality, but requires >10-hour procedures and specialized expertise available only at select institutions. Because treatment strategies continue to be debated nationally, and outcomes remain poor, there is a pressing unmet need for novel clinical decision support tools. We aim to develop two complementary modeling methods to support clinical decision making in CHD patients with PPS: 1) a mechanistic multiscale model of pulmonary fluid solid growth melding fluid structure interaction (FSI) and vascular growth and remodeling (G&R), and 2) a real-time uncertainty-aware digital twin model for virtual treatment planning to aid clinicians in identifying optimal treatment strategies. To accomplish these goals, we propose three specific aims: (1) Characterize mechanical and immunohistochemical properties of PA tissue in human PPS patients via biaxial testing and histology; (2) Develop and validate a computational modeling framework (melding hemodynamics and G&R) capable of predicting post-treatment hemodynamics in PPS; and (3) Develop and validate a fast, interactive Bayesian modeling framework for virtual treatment planning under uncertainty to aid near real-time clinical decision making for PPS, leveraging reduced order models. Our proposed study will tightly integrate modeling and experiments to improve physiological fidelity and clinical relevance of patient-specific models in an understudied patient population. The biaxial mechanical characterization of pediatric human PA tissue will provide much needed data on tissue properties in CHD which are currently absent from the literature. This project assembles an interdisciplinary team of engineers with expertise in hemodynamics modeling, cardiovascular biomechanics, mechanical characterization of biological tissues, and uncertainty quantification, and clinicians with expertise in pediatric cardiology/pulmonary vascular abnormalities, cardiothoracic surgery, and cardiac catheterization. Our translational objectives are to: (1) systematically compare treatment options for PPS and thus challenge the current standard of care and, (2) ultimately optimize treatments for individual patients, thus reducing unnecessary procedures and potential harm and improving long-term outcomes in this complex pediatric population.