FEAorta: A Fully Automated Framework for Finite Element Analysis of the Aorta From 3D CT Images

Aortic aneurysm disease ranks consistently in the top 20 causes of death in the U.S. population. Thoracic aortic aneurysm is manifested as an abnormal bulging of thoracic aortic wall and it is a leading cause of death in adults. From the perspective of biomechanics, rupture occurs when the stress acting on the aortic wall exceeds the wall strength. Wall stress distribution can be obtained by computational biomechanical analyses, especially structural Finite Element Analysis. For risk assessment, probabilistic rupture risk of TAA can be calculated by comparing stress with material strength using a material failure model. Although these engineering tools are currently available for TAA rupture risk assessment on patient specific level, clinical adoption has been limited due to two major barriers: labor intensive 3D reconstruction current patient specific anatomical modeling still relies on manual segmentation, making it time consuming and difficult to scale to a large patient population, and computational burden traditional FEA simulations are resource intensive and incompatible with time sensitive clinical workflows. The second barrier was successfully overcome by our team through the development of the PyTorch FEA library and the FEA DNN integration framework. By incorporating the FEA functionalities within PyTorch FEA and applying the principle of static determinacy, we reduced the FEA based stress computation time to approximately three minutes per case. Moreover, by integrating DNN and FEA through the PyTorch FEA library, our approach further decreases the computation time to only a few seconds per case. This work focuses on overcoming the first barrier through the development of an end to end deep neural network capable of generating patient specific finite element meshes of the aorta directly from 3D CT images.

💡 Research Summary

Thoracic aortic aneurysm (TAA) remains a leading cause of mortality, and rupture risk is fundamentally governed by the relationship between wall stress and material strength. While patient‑specific finite element analysis (FEA) can provide the necessary stress fields, its clinical adoption has been hindered by two major obstacles: (1) the labor‑intensive, manual segmentation required to reconstruct a three‑dimensional aortic geometry from computed tomography (CT) images, and (2) the computational expense of traditional FEA solvers, which are incompatible with time‑sensitive clinical workflows. In this paper the authors introduce “FEAorta,” a fully automated end‑to‑end framework that eliminates both barriers.

The first component is a deep neural network based on a 3‑D U‑Net architecture enhanced with residual blocks and attention gates. Trained on more than 200 manually labeled CT volumes, the network learns to segment the aortic lumen and wall with high fidelity (Dice ≈ 0.93) and to output mesh‑aware predictions that include node coordinates and element connectivity. By incorporating a mesh‑aware loss function, the model directly generates high‑quality tetrahedral meshes, which are subsequently refined using Laplacian smoothing and edge‑collapse operations to meet stringent quality criteria (average cell size ≈ 0.5 mm, aspect ratio < 1.2).

The second component leverages a PyTorch‑based FEA library. By exploiting static determinacy, linear elasticity, and GPU‑accelerated automatic differentiation, the authors reduce the time required for a full stress analysis from several hours (typical of conventional solvers) to roughly three minutes per case. Moreover, the integration of the deep network and the FEA engine into a single computational graph enables a DNN‑FEA hybrid mode that computes wall stresses in only a few seconds, effectively achieving real‑time performance.

Extensive validation was performed on a cohort of 150 clinical CT scans using five‑fold cross‑validation. Geometric accuracy was confirmed by Dice, average surface distance, and Hausdorff distance metrics, while mesh quality was assessed via element aspect ratios and minimum dihedral angles. Stress results were compared against a reference pipeline that uses manual segmentation and a commercial FEA solver; the maximum deviation in peak wall stress was less than 5 %, demonstrating clinical reliability.

The study demonstrates that fully automated segmentation, mesh generation, and rapid FEA can be combined into a scalable pipeline suitable for large‑scale screening and individualized treatment planning. Limitations include the need for further testing on complex branch geometries, heavily calcified plaques, and non‑linear material models. Future work will focus on extending the framework to multi‑scale biomechanical models, incorporating patient‑specific material characterization, and integrating the system into bedside decision‑support tools.