Project Grant K99AG103804

The National Institute on Aging awarded The General Hospital Corporation (Massachusetts General Hospital) $129,329 on August 5, 2026, under the Aging Research program (CFDA 93.866) to develop advanced artificial intelligence methods for computational laminar MRI to diagnose preclinical and prodromal Alzheimer's disease.

The project will create deep learning-based segmentation models to improve the resolution of in vivo structural MRI and enable accurate segmentation of the supragranular and infragranular cortical layers, hippocampal subregions and layers, and layer II of the entorhinal cortex. The applicant will integrate multi-modal high-resolution imaging data—including in vivo hippocampal scans, ex vivo MRI, OCT, histology, and light-sheet fluorescence microscopy—and embed prior neuroanatomical knowledge through Bayesian learning. Using longitudinal segmentation maps, the project will investigate early laminar degeneration in the medial temporal lobe during preclinical Alzheimer's disease and in perirhinal, parietal, and frontal cortical areas during prodromal stages. The approach is designed to discover novel laminar MRI biomarkers potentially 2–3 years earlier than conventional volumetric MRI atrophy measures, improving diagnostic accuracy for early Alzheimer's disease without additional cost.

The award funds the K99 phase, during which the applicant will develop the segmentation models while gaining new training in clinical staging and Alzheimer's disease neuropathology. Work is performed at Massachusetts General Hospital in Charlestown, Massachusetts. The period of performance runs from August 5, 2026, through July 31, 2028.