The University of Louisville, through its Office of Research & Innovation Division, received a $464,073 Project Grant award from the National Institute of Diabetes and Digestive and Kidney Diseases (CFDA 93.847) to develop a generative AI-based system for accurate prediction of deceased donor liver-transplant (DDLT) outcomes and viability.
The proposed system will use deep learning techniques to produce "virtual" Masson's trichrome stained slides from input hematoxylin and eosin frozen section slides of donor liver tissue. This will assist pathologists in evaluating allograft viability during the limited timeframe for transplantation. The system will also incorporate a predictive AI model to forecast post-transplant outcomes by fusing histopathological features from the generated and input slides with recipient clinical data. This technology aims to enhance the efficiency and accuracy of histopathological evaluation, improve post-transplant survival rates, reduce operational costs, and accelerate pathology workflows for DDLT procedures.
Generated 7/16/24, 7:09 AM