Alzheimer's disease (AD) is one of the most common causes of dementia in the world. Neurodegeneration, gliosis, and misfolded proteins such as amyloid plaques and tau tangles are neuropathological hallmarks in AD. In vivo imaging of these neuropathological lesions would be good biomarkers to understand pathophysiology as well as surrogate markers for clinical trials. We developed THK tau radiotracers including [18F]THK-5351 and tested them in humans. Validations studies identified monoamine oxidase-B (MAO-B) as the off-target binding substrate of [18F]THK-5351. Since the elevation of MAO-B, which is highly expressed in reactive astrocytes, were observed in various neurological conditions, MAO-B would be a promising target for imaging reactive astrogliosis. In fact, [18F]THK-5351 PET studies demonstrated that high tracer uptake in site susceptible regions to occur astrogliosis in various neurological disorders. However, the lack of binding selectivity affects the interpretation of PET images. Therefore, we performed lead optimization from [18F]THK-5351 generating a selective and reversible MAO-B PET tracer, [18F]SMBT-1. These translational and reverse translational studies, from the development of PET tracers to validation of PET images, led to the generation of new biomarkers. In this review, we will introduce the development of [18F]THK-5351, identification of off-target binding substrates, imaging-autopsy validations, new tracer development ([18F]SMBT-1), and finally recent clinical studies of [18F]SMBT-1.