Review: CT-based Computational Modeling to Identify High-Risk CAD
Key findings
- Consideration of the physiological characteristics of coronary plaque as well as anatomy may improve the diagnosis and management of coronary artery disease
- Computational fluid dynamics is a type of modeling based on medical imaging such as computed tomography angiography (CTA) that permits calculation of hemodynamic parameters, such as endothelial shear stress (ESS)
- CT-based computational modeling is also accepted as a standard method of measuring fractional flow reserve noninvasively (FFR-CT)
- Factors including ESS and FFR-CT have been shown to identify plaques that are at high risk of causing acute coronary syndromes
- Once the computational time is decreased, this technique should be useful for personalized medicine and to choose therapeutic interventions according to patient-specific plaque characteristics
Subscribe to the latest updates from Radiology Advances in Motion
Computed tomography angiography (CTA) can noninvasively assess atherosclerotic coronary plaque and provide information about the coronary tree, degree of stenosis and plaque morphology. However, those data alone contribute little to lesion-specific forecasting of a clinical event due to erosion or rupture. Combining anatomy with physiology may be considerably more powerful in predicting acute coronary events.
Technical advances in coronary CTA now allow assessment of hemodynamics via computational modeling. Researcher Parastou Eslami, PhD, and Udo Hoffmann, MD, cardiac radiologist of the Department of Radiology at Massachusetts General Hospital, and colleagues recently reviewed the role of coronary CTA and imaging-based computational modeling in identifying high-risk coronary lesions. Their paper appears in The International Journal of Cardiovascular Imaging.
Computational Fluid Dynamics
CT-based computational modeling relies on modeling blood flow and tissue behavior with mathematical equations. A prominent type of analysis is computational fluid dynamics (CFD), in which coronary blood velocity and pressure are computed using three-dimensional reconstructions of patient anatomy (based on imaging) and equations about fluid dynamics. Once the blood flow is known, hemodynamic parameters can be calculated, for example:
- Endothelial shear stress (ESS), which has been linked with atherogenesis, plaque progression and plaque vulnerability
- Axial plaque stress, which can occur at magnitudes 103 to 104 times higher than ESS and may be the biomechanical factor that contributes to plaque rupture
- Plaque structural stress, which at localized high levels can result in thrombosis and sudden ischemic events
Fractional Flow Reserve–CT
Fractional flow reserve (FFR) is the gold standard for evaluating whether a coronary lesion is significantly limiting blood flow to the myocardium. It is traditionally obtained invasively through a pressure wire during coronary catheterization. Now, software approved by the Food and Drug Administration (HeartFlow, Redwood City, CA) can provide measurements of ischemia from standard CTA. This physiologic measure is known as FFR-CT.
Data from the PROMISE trial showed that among patients with stable chest pain, FFR-CT ≤0.80 was significantly better than severe stenosis on CTA at predicting major adverse cardiac events. Similar results have been reported from other large clinical trials.
Acute Coronary Syndromes
In a recent CFD analysis published in JACC: Cardiovascular Imaging (n=72), ESS, FFR-CT and axial plaque stress identified plaque at high risk of causing acute coronary syndrome. In larger populations, it would be valuable to examine whether hemodynamic parameters can guide the management of coronary artery disease (CAD). For example, interaction of medical therapies such as statins with the hemodynamic milieu (e.g., ESS) could inform clinicians whether patients are candidates for medical therapy, revascularization or both.
Future Directions
Currently, computational modeling requires the expertise of trained engineers who reconstruct models from medical imaging, run simulations and extract the relevant data, a process that takes three to eight hours per patient. Researchers are working toward bedside applications that could provide real-time information. For instance, now that FFR-CT has been validated as a diagnostic alternative to invasive FFR, deep-learning networks are being trained to estimate FFR from CTA images.
view original journal article Subscription may be required
Learn about the Division of Cardiovascular Imaging
Refer a patient to Mass General Imaging