Radiology organizations to work together to promote AI use
The American College of Radiology and the Medical Image Computing and Computer Assisted Intervention Society have signed an agreement to advance the use of artificial intelligence, which the two groups announced at the recent NVIDIA GPU Technology Conference.
Mike Tilkin, ACR’s chief information officer and executive vice president, provided details on the agreement with MICCAI that promises to provide AI algorithms to facilitate medical imaging.
How significant is this memorandum of understanding (MOU) between your two organizations?
Tilkin: This new MOU is very significant. MICCAI promotes research and education and helps advance knowledge in areas such as AI and medical image computing. AI Imaging competitions have played an important role in advancing the progress of AI in imaging, and, among other things, MICCAI runs such competitions.
The ACR brings a strong clinical perspective, decades of experience creating imaging standards, and a history of promoting imaging informatics solutions, such as DICOM, that help the imaging technology landscape evolve and thrive. Working together, the organizations can help promote learning in a scientifically rigorous manner, target solutions that have the greatest clinical impact, and promote standards that encourage a useful clinical workflow.
What role will ACR play in this endeavor?
Tilkin: The ACR is actively creating use cases for imaging AI and will be working with MICCAI to leverage this knowledge base in their imaging AI competitions. ACR will also work with MICCAI to promote learning on a global scale.
How will the AI algorithms better meet the clinical needs of radiologists? What is the current state of affairs regarding this?
Tilkin: Although it’s still early, we believe AI algorithms will be useful in a variety of areas throughout the imaging lifecycle and will help radiologists be more efficient and provide better patient care. Radiology has played a leadership role in the application of advanced technology in medicine, and we believe AI represents another important area where advances in technology have the potential to improve clinical outcomes if applied in a safe and effective manner. Although advances in “General AI” that involve thinking computers are still well into the future, we do see “Narrow AI” solutions that target very specific clinical problems or workflow needs on the very near-term horizon.
Will the algorithms be made available to the public? How can they be accessed?
Tilkin: We believe algorithms will be integrated into clinical applications throughout the imaging life-cycle. At the ACR Data Science Institute, we are actively working on technically-oriented use cases (TOUCH-AI) which will help algorithm vendors identify and target areas that have the greatest clinical impact. We are also working on strategies to ensure appropriate validation pre-deployment (CERTIFY-AI) as well as ongoing monitoring while in the clinical setting (ASSESS-AI).
When will they be released?
Tilkin: The commercial application of AI is already finding its way into the market. However, this industry is still in a very, very early stage. ACR’s goal is to provide standards and an infrastructure for monitoring and assessment that will help development grow in a way that is safe, effective and has the maximum positive impact on patient care.
Will ACR and MICCAI continue to work together on other projects in the near or distant future?
Tilkin: Yes. We are already starting to plan how we can collaborate on imaging AI competitions so that we can leverage MICCAI’s energy and experience in this area with the ACR’s strong clinical expertise.