Implications and opportunities for AI implementation in diagnostic medical imaging formulated in workshop report published in the journal, Radiology
Radiologists train for years to attain the skills to interpret subtle and not-so-subtle distinctions in medical images. Artificial intelligence (AI) is poised to make profound impact on their efforts, assisting human experts with computer-powered algorithms to recognize anatomical anomalies, enhance interpretation, and improve classification of medical imaging results.
In August 2018, experts in medical imaging participated in a workshop at NIH in Bethesda, Md., to explore the future of AI in medical imaging. The workshop was organized by the National Institute of Biomedical Imaging and Bioengineering (NIBIB) and co-sponsored by NIH, the Radiological Society of North America (RSNA), the American College of Radiology (ACR), and the Academy for Radiology and Biomedical Imaging Research (The Academy). A foundational research roadmap for AI in medical imaging was published April 16, 2019, as a special report in the journal Radiology.
The collaborative report underscores the commitment by standards bodies, professional societies, governmental agencies, and private industry to work together to accomplish a set of shared goals in service of patients, who stand to benefit from the potential of AI to bring about innovative imaging technologies.
“Artificial intelligence applications promise to enhance every stage of medical imaging, from image creation, to augmenting image interpretation, to improving workflow and communication with non-imagers and patients,” said Kris Kandarpa, M.D., Ph.D., director of research sciences and strategic directions at NIBIB, and senior author of the reports. “It was essential for NIBIB to organize this workshop on behalf of the NIH and with eminent medical imaging groups to create a research roadmap that will enable deployment of AI advances in medical imaging across the nation.”
“The scientific challenges and opportunities of AI in medical Imaging are profound, but quite different from those facing AI generally,” said co-author, Curtis P. Langlotz, M.D., Ph.D. “Our goal was to provide a blueprint for professional societies, funding agencies, research labs, and everyone else working in the field to accelerate research toward AI innovations that benefit patients,” Dr. Langlotz is a professor of radiology and biomedical informatics, director of the Center for Artificial Intelligence in Medicine and Imaging, and associate chair for information systems in the Department of Radiology at Stanford University, and RSNA Board Liaison for Information Technology and Annual Meeting.
“We all appreciate NIBIB hosting this important event,” said Bibb Allen, MD, workshop co-chair and chief Medical Officer of the ACR Data Science Institute. “The workshop was a great opportunity for the radiology community to come together to discuss the needs and challenges for AI research facing our specialty and develop a roadmap for future foundational and translational research activity in medical imaging.”
“The workshop expanded our collective knowledge about the potential utility for Artificial Intelligence to improve the efficiency and accuracy of diagnostic systems,” said Steven E. Seltzer, M.D., FACR, Radiology Department Chair Emeritus, Brigham and Women’s Hospital; Distinguished Cook Professor of Radiology, Harvard Medical School; and Health and Science Policy Fellow of The Academy. “If the need for precision diagnosis in the future requires collation of images from radiology, pathology and ‘omics’ systems into a diagnostic cockpit, the human observer will need considerable help from computers to extract optimum information from multiple, disparate sources. AI can be a key ingredient in this process.”
The report describes innovations that would help to produce more publicly available, validated and reusable data sets against which to evaluate new algorithms and techniques, noting that to be useful for machine learning these data sets require methods to rapidly create labeled or annotated imaging data. The roadmap of priorities for AI in medical imaging research includes:
- new image reconstruction methods that efficiently produce images suitable for human interpretation from source data,
- automated image labeling and annotation methods, including information extraction from the imaging report, electronic phenotyping, and prospective structured image reporting,• new machine learning methods for clinical imaging data, such as tailored, pre-trained model architectures, and distributed machine learning methods,
- machine learning methods that can explain the advice they provide to human users (so-called explainable artificial intelligence), and
- validated methods for image de-identification and data sharing to facilitate wide availability of clinical imaging data sets.
Article reference: Langlotz, CP, et al. A Roadmap for Foundational Research on Artificial Intelligence in Medical Imaging: From the 2018 NIH/RSNA/ACR/The Academy Workshop. Radiology. April 16, 2019.