Clinical classification of thoracic aortic morphology informed by dynamical modeling of anatomic variables from CT data

Anatomic descriptors were extracted from CT angiography (CTA) images to create a morphologically defined training data set of patient disease state. The sparse identification of nonlinear dynamics (SINDy) framework was employed to identify the temporal equations that describe the dynamic evolution of aortic structure over the ensemble of time points for each patient. This data-driven approach sparsely selects terms from a comprehensive library of candidate functions, creating parametric models that capture the essential dynamics of each postoperative aortic pathway. Patient group dynamics were organized solely from temporal morphological characteristics and were then directly compared to clinical categorizations of patient outcomes. This approach will provide a data-driven tool for modeling aortic morphology changes of patients receiving endovascular repair and informing clinical stratification, enhancing our understanding of the biomechanical response of the aorta post-intervention.