GTC-DC 2019: NIH-NVIDIA Cancer Ecosystem AI Pilot
Gregg Cohen, NIH Clinical Center; Brad Wood, NIH
NIH & NVIDIA have set up a public-private partnership (CRADA) to explore and deploy AI tools and approaches in cancer imaging. Efforts in cancer AI will have a foundation in medical imaging, but will attempt to integrate digital pathology and molecular analysis as well into the interrogation and characterization of cancer, as well as the use of AI in clinical trials, drug discovery, and predictive models of outcomes. Early efforts will focus on liver, prostate, and kidney cancer, as well as exploring ways to integrate AI tools into the clinical radiology workflow. Areas of particular interest are AI model implementation on PACS, image processing workstations, and for interventional tools such as segmentation and registration required for multi-modality fusion guided interventions. As America’s Research Hospital, the NIH Clinical Center is the world’s largest clinical research hospital, and is an ideal milieu for the deployment and integration of AImodels into clinical workflows in cancer imaging. Radiology and Imaging Sciences, the Center for Interventional Oncology, and the Molecular Imaging Branch (in the National Cancer Institute) are partnering with NVIDIA to demonstrate the power and clinical applications of deep learning AI tools within a clinical Radiology Department.We will discuss current projects under development as well as examples of clinical uses of AI in cancer imaging. We will also discuss the mechanics and configuration of our systems, including an overviews of the data center environment, data management, data governance, workflows and workflow management, security and system integrity, with a system centered around an on-site GPU environment.