PROJECT SUMMARY Since its discovery, adult mammalian hippocampal neurogenesis—particularly human—has attracted attention and controversy. Envisioned as a unique, intrinsic capacity of the brain center for learning and memory and mood control to repair and regenerate, over the past three decades it has been scrutinized in model organisms, which provided a wealth of data confirming its functional relevance. The case for human neurogenesis, however, has faced a much harder road to acceptance because the only means to study it has been by immunostaining of the postmortem tissue. It is thus of no surprise that we know very little about it. While most agree that adult human hippocampus harbors newborn neurons that decline with age and with diseases such as Alzheimer’s, their functional importance has proved elusive without a live and non-invasive measure. As noted in the RFA, developing a means of measuring neurogenesis rates in vivo in a non-invasive manner is of critical interest. Doing so in a technique that can be readily used in not only animal models, but also in humans would give us a vital tool in our effort to not only understand neurogenesis per se, but to also understand how neurogenesis may lead to impairments found in aging and in AD/ADRD. This proposal takes as its foundation the innovation by our group of the first and only in vivo magnetic resonance spectroscopy (MRS)-based marker of neurogenesis: a lipid-based signal resonating at 1.28 ppm. While our group’s initial work in this area has identified and validated this biomarker, several critical gaps exist that must be addressed before it can be widely adopted in human studies of aging and AD/ADRD. Our goal in this proposal is to adapt a number of existing techniques to the accurate measurement and quantification of the 1.28ppm signal and to further develop analytical methods based on deep machine learning so that it can be broadly and reliably used to assess levels of and changes in human adult neurogenesis. In particular, we will: 1) Acquire data from phantoms, in vivo and ex vivo mice, and humans using techniques that will allow for more reliable quantification and validation, 2) Adapt the MEGA-PRESS technique successfully used to quantify GABA to isolate the neurogenic-associated signal at 1.28ppm from the nearby lactate signal; 3) Adapt pre-processing tools we have developed in related studies that enhance signal-to-noise in MRS signals from the hippocampus; 4) Further develop time-domain based processing tools to isolate the neurogenic signal from overlapping components; and 5) Further develop a neural-network based approach to detect and quantify it. In each of these areas, we have existing solutions that are functional, but we believe can be improved upon to provide more reliable and robust quantification of the neurogenic signal. Here, we will formally evaluate the new approaches relative to the existing approaches to produce a final acquisition-through-quantification pipeline ...