A collection of de-identified annotated medical imaging datasets
Artificial intelligence (AI) and related technologies are increasingly focused on healthcare, particularly imaging. These technologies have the potential to transform many aspects of patient care in several areas of healthcare, and in particular medical imaging. There are many early research studies suggesting that AI can perform as well as or better than human experts at key healthcare tasks, such as diagnosing disease. However a key limitation that persists is the availability of high-quality labeled medical data for research and education use. Biostatistical applications, machine learning, deep learning, and causal analyses are all most successful with large-scale data beyond what is available in any given organization.The Stanford Center for Artificial Intelligence in Medicine and Imaging (AIMI Center’s) mission is to solve clinically important problems in medicine using AI by developing and supporting transformative medicine AI applications and the latest in applied computational and biomedical imaging research to advance health. While early efforts by the Stanford Center for Artificial Intelligence in Medicine and Imaging (AIMI) have focused on infrastructure to responsibly make medical imaging datasets open to the global research and education communities for specific use cases, more work is needed to create a “Medical ImageNet” ecosystem that affords larger, more diverse, more discoverable, and more feature rich datasets for medical imaging applications - and critically - convenient compute and cloud partners able to catalyze these biomedical research opportunities. Historically, logistical barriers to this have frustrated larger efforts.With the continued growth of cloud computing bridging the computing gap, an opportunity exists to transform population health and realize the potential of AI. Toward that goal we are proud to announce the Stanford AIMI - Microsoft “Medical ImageNet Project” as an open science medical imaging data community. Stanford AIMI and Microsoft will together harmonize de-identified annotated medical imaging data across organizations to foster transparent and reproducible collaborative research. The ultimate goal is to complete a petabyte-scale searchable repository of annotated de-identified medical images (radiology, echocardiography, ophthalmology, dermatology, pathology, and more) that are annotated and, eventually, linked to genomic data and electronic medical record information, for use in rapid creation of computer vision systems. We expect this initiative to amplify the Stanford AIMI work in enabling collaboration among clinicians, researchers, and data scientists focused on clinically important medical imaging problems. With this partnership and program together we expect to accelerate important research insights and discovery amongst researchers, institutions and companies. This project has the highest of ambitions in mind in opening high quality medical data to the research and education communities in the hopes of better health outcomes for all.