Multi-modality evaluation of high-risk coronary atherosclerotic plaque Summary Non-invasive, quantitative assessment of coronary atherosclerosis will lead to improved, personalized patient management for the leading cause of death in the US — cardiovascular disease. Detecting high-risk lesions at the earliest stages of coronary artery disease would facilitate timely medical interventions to hinder the pro- gression of coronary atherosclerosis and prevent catastrophic complications. Intravascular imaging modalities such as intravascular optical coherence tomography (IVOCT) have been used to identify the presence and characteristics of coronary atherosclerosis. IVOCT, with its high resolution and contrast, is recognized as the best method for identifying local high-risk lesions (e.g., thin cap fibroatheroma). However, intravascular imag- ing is invasive, limiting its applicability, especially for patients early in the disease process. Coronary computed tomography angiography (CCTA) is the only non-invasive imaging modality allowing the assessment of luminal stenosis as well as plaque morphology. We will develop new AI methods for CCTA evaluations of high-risk coronary atherosclerotic plaque by comparing them to concurrently-acquired, high-resolution/contrast IVOCT, deemed the best method to assess high-risk plaque. To enable a new, non-invasive evaluation of atheroscle- rosis, we will register CCTA images to concurrently-acquired IVOCT images and determine image features in CCTA that associate with and predict high-risk plaques as seen in IVOCT. In addition to this concurrent deter- mination of high-risk plaque in CCTA, we will take what we learn and apply it to the long-term prediction of ma- jor adverse cardiovascular events (MACE). Specifically, we will: 1) Create highly automated methods for as- sessing high-risk plaques seen microscopically in IVOCT and develop CCTA feature candidates suggestive of high-risk plaque; 2) Use novel IVOCT to CCTA registration to associate segmental CCTA features to IVOCT- defined high-risk plaque features and to create a CCTA classification model for high-risk plaque; and 3) Apply the most promising segmental CCTA features to predict long-term adverse outcomes in CCTA data. With suc- cess, our research will lead to decision support software for the prediction of MACE, facilitating revasculariza- tion strategies, therapeutic decisions, and furthering precision medicine approaches. The project team will build on expertise in cardiology, machine learning, biostatistics, and advanced image analysis of IVOCT and CCTA. In this project, I will build upon my experience in machine/deep learning analysis of intravascular images to include new training in image analytics of CCTA data, biostatistics, and bioinformatics of metabolomics and genomics, providing me with a foundation for a future career in cardiovascular disease. For example, I will be well situated to integrate CCTA image analytics and genomics to further my und...