Mendy Lebowitz
BME PhD Proposal Presentation

Date: 2026-12-07
Time: 4:00 pm
Location / Meeting Link: HSRB2-N657; https://zoom.us/j/95434878143

Committee Members:
Anant Madabhushi, PhD (Advisor); Hanjoong Jo, PhD; Lakshmi Prasad Dasi, PhD; Michael Girard PhD; Sadeer Al-Kindi, MD


Title: Multimodal Imaging Biomarkers of Cardiovascular Biological Aging

Abstract:
Cardiovascular (CV) disease is the leading cause of death and can progress silently for years before clinical events occur. Risk equations based on chronological age and conventional clinical factors cannot fully capture accumulated CV injury among people with similar profiles. People classified as low or borderline risk may still experience major adverse cardiovascular events (MACE), revealing risk heterogeneity that current approaches do not fully resolve. CV biological aging (CV-Age) is a multidimensional process shaped by intrinsic biology and lifetime environmental and psychosocial exposures. It may be reflected in cumulative subclinical injury, inflammation, fibrosis, and vascular remodeling. In this proposal, CV-Age is operationalized as an imaging-derived age estimate based on age-associated features quantified from medical images. Routine chest computed tomography (CT) and retinal photography offer complementary views of cardiovascular biology: CT captures macrovascular, structural, and cardiometabolic features, whereas retinal photographs visualize microvascular structure. Explainable artificial intelligence (AI) can quantify aging patterns from images already acquired in clinical care. The objective is to develop and evaluate CT-derived CV-Age and retinal-derived CV-Age independently, then determine whether their integration adds information beyond either modality alone. The central hypothesis is that these measures capture distinct, potentially complementary manifestations of cardiovascular aging that are associated with MACE. Aim 1 will develop CT-derived CV-Age from chest CT radiomics and evaluate its association with future MACE. Aim 2 will develop retinal-derived CV-Age from fundus oculomics and evaluate its association with future MACE. Aim 3 will test whether integrating the independently evaluated measures improves age estimation or MACE risk stratification in cohorts with linked CT and retinal imaging. This proposal will determine whether either modality or their combination adds incremental clinical value to existing risk algorithms. By transforming underused routine-image data into an individualized description of cardiovascular aging, findings from this proposal could support earlier and more personalized risk recognition.