Fully automated construction of three-dimensional finite element simulations from Optical Coherence Tomography

TL;DR

The paper tackles the challenge of predicting atherosclerotic plaque stability by generating patient-specific 3D finite element models from Optical Coherence Tomography (OCT) images. It introduces a fully automated pipeline that starts from CNN-based tissue labeling of labeled OCT frames, uses signed distance function interpolation to create a 3D isotropic voxelgrid, and applies automated meshing and boundary-condition assignment to produce ABAQUS-ready inputs. Key contributions include automated, high-fidelity 3D reconstructions that capture fine calcifications, validated mesh convergence with an average strain error of $1.61\%$ and an average stress error of $5.9\%$, and a workflow that reduces per-pullback model creation to minutes, enabling large-scale analyses. The approach enables mechanically informed lesion assessment and has the potential to support patient-specific diagnoses, treatment planning, and virtual interventions in coronary artery disease.

Abstract

Despite recent advances in diagnosis and treatment, atherosclerotic coronary artery diseases remain a leading cause of death worldwide. Various imaging modalities and metrics can detect lesions and predict patients at risk; however, identifying unstable lesions is still difficult. Current techniques cannot fully capture the complex morphology-modulated mechanical responses that affect plaque stability, leading to catastrophic failure and mute the benefit of device and drug interventions. Finite Element (FE) simulations utilizing intravascular imaging OCT (Optical Coherence Tomography) are effective in defining physiological stress distributions. However, creating 3D FE simulations of coronary arteries from OCT images is challenging to fully automate given OCT frame sparsity, limited material contrast, and restricted penetration depth. To address such limitations, we developed an algorithmic approach to automatically produce 3D FE-ready digital twins from labeled OCT images. The 3D models are anatomically faithful and recapitulate mechanically relevant tissue lesion components, automatically producing morphologies structurally similar to manually constructed models whilst including more minute details. A mesh convergence study highlighted the ability to reach stress and strain convergence with average errors of just 5.9% and 1.6% respectively in comparison to FE models with approximately twice the number of elements in areas of refinement. Such an automated procedure will enable analysis of large clinical cohorts at a previously unattainable scale and opens the possibility for in-silico methods for patient specific diagnoses and treatment planning for coronary artery disease.

Figures (5)

• Figure 1: The automated framework pipeline. A CNN developed by OlAt18 segments the raw OCT images into material labels corresponding to several tissue classes. Image processing steps modify the CNN labeled OCT frames to create anatomical geometry, followed by the signed distance function (SDF) based interpolation procedure. Healthy artery tissue, lipid and calcium are represented by the dark gray, blue, and light gray colors.. Refinement layers (gold and green) are then created to allow for extra refinement at sites of expected high stress. The golden layers are refined the most followed by the layer in green. Meshing steps utilizing open-source software create the volumetric mesh with the applied boundary conditions and material properties. The FE ABAQUS file is then ready to be used in a simulation.
• Figure 2: Illustration of the frames that were investigated for the mesh convergence study. Frames shown in (a) consist of complex geometry that includes large contents of lipid (blue) and calcium (light gray) embedded within arterial tissue (dark gray). Region highlighted shows the material distribution (b) and tissue stress (c). The arterial stress is significantly concentrated at the region of the thinnest fibrous cap. Stress of calcium was excluded to highlight arterial stress distribution. Peak stress is highlighted by the (*) marker
• Figure 3: Comparison of the generated 3D geometry created by the semi-automated CAD based method employed by Ka21 (left) vs the newly proposed method in this paper (right). The semi-automated method, that relied on manual lofting using CAD software, produces a model of similar morphology for healthy arterial tissue (dark gray), lipid (blue) and calcium(light gray). A noticeable difference is that the automated method includes some extra smaller, but mechanically significant calcifications (black arrows), otherwise unfeasible to include in the manual process.
• Figure 4: Cross-sectional view of eight different sections from four separate OCT pullbacks, each exhibiting a distinct morphology, resulting in a variety of stress responses. The peak stress is represented by a (*) marker.
• Figure 5: View of the inner vessel wall surface of healthy arterial tissue with embedded lipid and calcium volumes. The positioning of the lipid and calcium volumes significantly impacts the surface stress of the artery. The large deposit of lipid that resides near the surface of the lumen (a), induces a region of higher stress concentration on the arterial surface (b). Although the volume of clarifications may be small in region (c), the stress inducing effects of a stiff localized deposit are significant (d).