Karim Kadry, MA
Ross Straughan, Karim Kadry, Elazer R. Edelman, Farhad R. Nezami
Farhad R. Nezami
Research Category: Cardiovascular, Diabetes, and Metabolic Disorders
Coronary artery disease persists as the world’s leading cause of death. Stability of atherosclerotic lesions are dependent on both the material and morphological properties of the arterial constituents, with excessive plaque structure stress (PSS) instigating rupture. Studies utilizing Finite Element Analysis (FEA) reveal that current clinical imaging modalities and metrics do not fully capture the complex 3D morphological-mechanical response. Although FEA has procured a deeper understanding of the biomechanics through optical coherence tomography (OCT), modality limitations such as frame sparsity and material contrast have prevented from large-scale automatized studies to be performed and are often limited to inferior 2D simulations. Herein, we have produced a fully automated software platform to analyze the stress distributions from deep-learning labeled OCT images by interpolating labeled frames and subsequently producing 3D FEA-ready simulations. A preliminary study highlighted that frames with a high PSS (over 150 kPa) were founded to have significantly larger lipid arc angles, lumen-plaque eccentricity, lipid area, and calcium area, mirroring clinical observations. Such promising results validate the potential for the platform to be utilized for plethora of applications from unprecedented large scale in-silico studies to enhance clinical metrics, optimization of patient-specific angiography operations, and training data for emerging AI simulations.
Excessive plaque formation in the arteries of the heart can potentially lead to rupture resulting in a heart attack or even heart failure. Plaques may rupture if the stress in the tissue gets dangerously high due to the material properties and arrangement in and around the plaque. Currently, optical coherence tomography (OCT) is the gold-standard method to visualize these materials and how they are arranged; however, computational simulations that predict how stress is distributed in the tissue shows that many of the metrics currently used may not be an accurate representation of the risk to a patient. Unfortunately, these simulations are often oversimplified and require days to create as they are not automated. Therefore, we have developed software that produces these structural simulations automatically in less than 15 minutes. In our preliminary study, clinical metrics that highlighted increased patient risk were also seen to have high structural stress from our simulations, illustrating how the developed software can be used as a clinical tool to identify patients at higher risk of plaque rupture. As these simulations can be produced in much less time, there is potential to examine large portions of the population to validate and improve current clinical metrics.