The Future of AI in Radiology
The Future of AI in Radiology
Dr. Eliot Siegel, Vice Chair of Imaging Informatics at the University of Maryland School of Medicine, joined Rad AI to discuss the future of AI in radiology. Dr. Sonia Gupta, AI thought leader and abdominal radiologist, moderated the discussion.
Dr. Gupta: Thank you to everyone who tuned into our webinar. I think it's really exciting for us to have this great turnout and shows how much engagement and excitement there is about AI in radiology.
Dr. Siegel began the discussion by describing where artificial intelligence will have the greatest impact in the near term.
Dr. Siegel: For me, the most important thing that AI can be applied to is actually the radiology report itself. So many developments in AI and natural language processing are allowing us to rethink the way we generate radiology reports and how we can do that more productively and efficiently. There’s incredible potential that we are not currently realizing to be able to significantly improve our efficiency and to reduce burnout as well.
Dr. Chang spoke about the impact Rad AI is having now in over 15% of the radiology market.
Dr. Chang: Rad AI Omni automatically generates the impression section of the radiology report, customized to each individual radiologist. It also handles automatic consensus guideline recommendations, helps with the automation of MIPS tracking (ensuring that the correct MIPS language is included in reports and improving billing language), reduces the need for addenda in reports, and provides error checking as well. We help radiologists by catching voice recognition typos, and other types of clinical errors in the report.
So this really saves radiologists time, reduces radiologist burnout, and improves the consistency and accuracy of reports. And because it gives you a little bit of a break after each complicated report, it makes people less stressed out -- it gives you a breather, so it reduces fatigue over the course of a day.
Dr. Gupta asked about how artificial intelligence can help serve as a virtual assistant for radiologists and reduce their burnout.
Dr. Gupta: At the same time that we have this increase in volume from the backlog of COVID-19, we also have decreasing reimbursement. And so unfortunately our survey data shows that 88% of US radiologists feel that they are at or over capacity and 25% of them say they're just unhappy being a radiologist.
Dr. Chang: When you're working really busy nights it's hard to always keep track of the incidental findings, always remember the different consensus guidelines and the latest recommendations, and be able to apply those appropriately at all times to create the best possible report.
And so at Rad AI what we really tried to focus on is making sure to help the radiologist to be able to reduce the amount of cognitive load on every study so that we can automatically remember those incidental findings for you -- remember the consensus guideline recommendations to put them in automatically. That really helps alleviate some of the pressure from this high impact, non-stop imaging volume that we have on overnight shifts.
Finally Dr. Siegel and Dr. Chang spoke about how artificial intelligence can help in closing the loop on significant incidental findings.
Dr. Siegel: We make so many findings, and so many recommendations, and so many suggestions for follow-up -- and yet there's not a mechanism to be able to track those.
Dr. Chang: Rad AI Continuity works with radiology groups and health systems on the automatic communication and tracking of recommendations for incidental findings in order to close the loop and ensure that the appropriate follow-ups are being done. We tend to find that about 15% of all radiology reports have significant incidental findings that need follow-up, yet about 60% of those don't have specific follow-up recommendations, or don’t even mention the significant incidental finding in the impression.