Development of physiologically-informed computational coronary artery plaques for use in virtual imaging trials

Area of Science:

• Biomedical imaging
• Computational modeling
• Cardiovascular disease research

Background:

• Cardiovascular disease (CVD) is a leading global cause of death, with coronary artery disease (CAD) being a major contributor.
• Traditional clinical trials for validating cardiac imaging technologies are time-consuming, expensive, and require large patient cohorts.
• Virtual imaging trials (VITs) offer a promising alternative by using virtual patients and accurate imaging device models.

Purpose of the Study:

• To develop physiologically-informed, realistic coronary plaque models for cardiac imaging VITs.
• To create anatomically variable plaque models with clinical realism for simulation purposes.

Main Methods:

• A deep convolutional generative adversarial network (DC-GAN) was trained on high-resolution histology images to generate plaque models.
• Finite element analysis (FEA) was used to assess the stability of the generated plaque models under physiological conditions.
• The plaque models were integrated with the XCAT computational phantom for simulations comparing energy-integrating detector (EID) CT and photon-counting detector (PCD) CT.

Main Results:

• The DC-GAN successfully generated realistic and anatomically diverse coronary plaque models.
• Simulation results demonstrated the utility of these models for comparing different cardiac imaging devices.

Conclusions:

• Generated CAD pathologies can be incorporated into computational phantoms for VITs.
• These models provide a "known truth" for quantitative optimization and evaluation of cardiac imaging technologies.