Doctor of Philosophy (PhD)



Document Type



Positron emission tomography (PET) plays a crucial role in neuroimaging, particularly in the diagnosis and study of Alzheimer's disease (AD). However, limitations remain in its widespread application, including the cost and availability of specific radiotracers, such as tau PET. This dissertation addresses these challenges through a two-pronged approach, focusing on standardized data processing and the development of novel analysis techniques. In Chapter 2, we investigate the feasibility of replicating a standardized brain PET processing pipeline established by a large, flagship research consortium in a single laboratory setting. This chapter examines the practicality and potential challenges associated with implementing such a pipeline, emphasizing the need for improved resources and tools to ensure reliable and reproducible PET analysis. In Chapter 3, we propose a novel approach utilizing deep learning to overcome limitations in tau PET scan accessibility and cost. This chapter introduces a multi-channel generative adversarial network (MC-GAN) capable of synthesizing tau PET images based on readily available and more cost-effective neuroimaging modalities like fluorodeoxyglucose (FDG) PET, amyloid PET, and MRI. This method has the potential to significantly expand access to tau PET data, enabling broader assessment of tau burden and identification of individuals at higher risk of developing AD. In Chapter 4, we apply the proposed MC-GAN architecture to real-world data from a middle-aged cohort to investigates the connection between lifespan vascular exposures and mid-life brain tau accumulation. This chapter analyzes the association between longitudinal cardiometabolic health data and tau PET measures. The findings suggest that higher body mass index (BMI) from early-adulthood to midlife, and early-adulthood diastolic blood pressure (DBP), are associated with greater tau deposition, highlighting potential links between vascular health and AD pathology.



Committee Chair

Carmichael, Owen

Available for download on Wednesday, April 02, 2031