ABSTRACT Early detection of tuberculosis (TB) disease to prevent transmission and disability is hampered by the lack of precise biomarkers to diagnose and predict TB progression. The objective of this study is to develop computational imaging tools to characterize and diagnose individuals without TB symptoms who will progress to sputum culture-positive or symptomatic TB disease. In Aim 1, we will combine computed tomography (CT) scans from existing and prospective TB household contact studies to derive a high-resolution radiomic signature to predict and characterize early disease pathology. In Aim 2, we will evaluate methods to enhance performance of field-deployable chest-X-ray computer aided detection (CAD) systems to diagnose early TB using transfer learning, image-to-image training, and integration of clinical and epidemiologic variables. Our goal is to develop a fundamental toolbox for research, diagnosis, and targeted preventive treatment of early TB to prevent transmission and accelerate TB control.