Bradley Leshnower and Wei Sun Named Co-Investigators of CTSA Pilot Grant
In the recent round of awards announced by the Georgia Clinical and Translational Science Alliance's (CTSA) Pilot Translational & Clinical Studies (PTCS) program, Bradley Leshnower, MD, assistant professor of surgery, Emory Division of Cardiothoracic Surgery, and Wei Sun, PhD, associate professor, Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University, received a $39,450 award as co-investigators for their study, "Estimation of In Vivo Mechanical Properties of the Aortic Wall from Cardiac Images: A Pilot Study."
The Georgia CTSA is a strategic alliance of the Emory University School of Medicine, Morehouse School of Medicine, Georgia Institute of Technology, and the University of Georgia that concentrates basic, translational, and clinical research investigators, community clinicians, professional societies, and industry collaborators in dynamic clinical and translational research projects. The PTCS program is considered a catalyst and vehicle for the transformation of clinical and translational science in Georgia, and promotes new networks of multidisciplinary and inter-institutional research teams to re-engineer Atlanta's health sciences enterprise.
Dr. Leshnower and Dr. Sun's project will endeavor to develop a validated process initiated by 3D CT imaging data that could accurately and quickly predict in vivo mechanical properties and the individualized risk of rupture or dissection in patients with thoracic aortic aneurysms (TAA). Such a system could provide additional information regarding indications for elective surgical repair, which would fulfill a critical need, as TAA—an abnormal widening or ballooning of the aorta due to weakness in the vessel wall—is highly lethal, carries a five-year survival rate of 54 percent in patients left untreated, and is often asymptomatic, with death being the first symptom for 95 percent of patients.
Standard imaging approaches such as magnetic resonance imaging (MRI) and echocardiography can only measure the existing status of the thoracic aorta, and are unable to predict the risk of such events as rupture under elevated arterial pressures. Drs. Leshnower and Sun will seek to design an algorithmic system powered by machine learning (MN) and deep learning (DN) interpretive tactics that can analyze patient-specific 3D CT imaging data and yield fast and dependable analysis for clinical diagnosis and treatment planning. MN and DN are branches of computer science wherein computer systems can be designed that have the ability to "learn" progressively from the gradual accretion of data. Such systems can evolve to the degree that they can detect patterns, connections, and cause and effect relationships, and then make data-driven predictions and recommendations.
The investigators plan to capture heterogeneous elastic mechanical properties of the thoracic aorta—essentially measurements of the degrees of expansion and contraction of the arterial walls—and the reactions of these properties to varying levels of strain and pressure to build an analytical tool that can accurately define, classify, and estimate the elastic mechanical condition of particular patients' aortic walls from clinical 3D CT images. Both porcine aorta tissue and human patient data will be used to validate the tool. From this tool, a second, DL-based model housed in a multi-GPU server will be developed that will ideally be capable of conducting rupture risk analyses by identifying material properties from in vivo patient aorta geometries imaged with CT scanning within seconds, providing fast feedback for clinical diagnoses and decision making.
"If successful, these methods will enable individualized computational assessment of the condition of the aorta and enhance our ability to make individualized recommendations for the treatment of patients with TAA," says Dr. Leshnower.