Quantification of the rate of CSF clearance from PET scan images as a new computational marker for Alzheimer's Disease

The current diagnostic techniques for Alzheimer's Disease rely on imaging the brain using CT, MRI, or PET. These techniques are typically subjective as they heavily rely on human judgment. This study attempts to introduce a more systematic approach that relies on the strong relationship between the rate of cerebrospinal fluid (CSF) clearance and the risk of Alzheimer's Disease. More specifically, this study investigates the possibility of extracting CSF flow rate from PET scan images by solving an inverse computational problem. The idea is to construct a model with unknown clearance rates as the model parameters, and then solve an optimization problem where those unknown parameters are tuned so that the model produces contrasts similar to those observed in the PET scan. In a blind study, we compare calculated CSF flow rates obtained from this process with the normal ranges for healthy patients to quantify the risk of Alzheimer's Disease. This new marker is then compared with the standard practice involving physicians' medical opinion that is directly based on the brain images.