Whereas amyloid deposition tends to occur broadly throughout cortical regions and with relatively uniform pattern of progression over time, the distribution of tau is highly focal at the early stages of pathology and its propagation to other regions demonstrates a more complex spatiotemporal evolution. In addition the relatively lower abundance in brain of tau compared to amyloid imparts greater requirements for binding affinity and selectivity of tau radiotracers. Although second generation tau tracers reported within the past few years appear to have more favorable properties in this regard, all tau tracers studied to date have been reported to exhibit off-target binding to various substrates that can confound the quantification and interpretation of imaging data. The goal of this application is to develop innovative technologies that improve the quantification of tau PET imaging and enhance its utility in both research and eventual clinical settings. To address the focal nature of tau deposition especially at prodromal stages of disease, we propose a novel kernel-based reconstruction method that uses structural MR images as prior information and incorporates motion correction to obtain images with unprecedented spatial resolution, allowing improved localization and quantification of tau in small structures such as the rhinal cortex and the locus coeruleus. To mitigate the effects of off-target binding, we propose a kinetic analysis strategies to identify non-specific, tau-specific, and off-target components. We also propose modeling tactics to estimate maps of cerebral perfusion and to complement measures of tau burden as an additional index of brain function. Lastly, we propose deep learning strategies to gain new insights into the spreading of tau and to predict this progression in individual subjects.