By Mark Casner, RadSite Accreditation Committee Chair
Artificial Intelligence (AI) is a new buzz phrase that is starting to pervade every industry. We now read about self-driving cars, for instance, and marvel at the abundance of sensors that control their every movement. On-board computers synthesize data to ensure the car operates safely. We still hear of accidents, however, proving that more development needs to be done to transition from driver to driverless technology.
It’s the same with medicine. An early precursor in radiology was the introduction of digital mammography many years ago. Technology evolved from there to support programs that could understand a physician’s speech patterns, making dictation much faster and easier. Now AI holds promise for even greater leaps forward in imaging.
How can AI make a difference? One of its key strengths is in synthesizing the overwhelming amount of data that is generated, a marked advantage in an era of information overload. Eliot Siegel, MD, RadSite’s Standards Committee Chair, notes, “The real disruption is going to come in the way that people consume algorithms. In general, we are constrained to pick a particular PACS system from a particular vendor. But with an increasing number of really interesting and creative AI software algorithms under development, we’ll be able to consume data with a system or engine that delivers best-of-breed algorithms from different vendors. That will be very disruptive in radiology—in a positive way.”
In terms of the practice of radiology, AI will not replace the radiologist or the techs, but it may allow for more efficient and quicker interpretations. The ability to use computer-based algorithms to find undiagnosed tumors and other medical conditions holds the potential to reduce errors and improve clinical outcomes.
That in turn points to AI’s ability to find identify evidence-based interventions and promote broader population health strategies. “Two areas where AI is working today is in improving MRI and PET scan image processing,” comments Siegel. “AI allows for major reductions in time of imaging and in radiation dose for PET and contrast dose for MRI. AI is also being used in the United States to triage patients with suspected stroke who have had head CT scans to the top of a reading worklist. It’s also being used to notify clinicians and radiologists about suspicious studies that have been completed but not yet interpreted.”
As in the auto industry, more work needs to be done to fully realize AI’s potential in imaging.
And how will AI impact the accreditation process? Accreditation Organizations (AOs) must stay on the cutting edge as this technology continues to evolve. Stay tuned as RadSite will have several important announcements in 2019 on how we are leveraging technology to improve the quality assessment process for advanced diagnostic imaging.