Trends in the adoption and integration of AI into radiology workflows

 

The integration of artificial intelligence into radiology PACS and enterprise imaging systems has become a big topic of discussion with IT vendors over the past couple years. This has become a bigger question from hospitals and radiology groups as there are now about 400 radiology related AI algorithms that have U.S. Food and Drug Administration clearance.

Amy Thompson, a senior analyst at Signify Research, is monitoring AI trends in radiology and said these algorithms are still relatively new to the market. Vendors are taking different approaches to how they integrate them into their radiology IT workflows. 

"We know, to maximize that value of AI, it needs to be embossed into the workflow. It needs to have that seamless flow and you don't want to open a pop-up window to have the AI results; you want it all within that single user interface," Thompson explained. 

However, since there are no standards set for how AI interfaces with various IT systems, she said it will take time. This is why many radiology IT PACS vendors are now partnering with a set number of AI companies that they feel offer a good product and work together to enable these types of seamless integrations. 

"To do this for 200-plus vendors is not realistic for any vendor to do, so there are still a lot of unknowns for IT vendors. Like, which AI vendor should I partner with, because you have 200-plus, and most of them are limited to three or four applications in terms of mammography, lung, brain and a few others. There is a lot of repetition currently in the market and there is not a lot of consolidation yet to have a top 10 or top 20 clear winners," Thomson said. 

Even asking radiology end users what vendors they prefer, there is no clear answer because 10 hospitals will give 10 different answers, she said.

AI in radiology faces the dilemma of interest vs demand

Vendors report there is a large amount of interest by radiologists in AI, but the actual number of purchases made has been much lower. Thompson said this is due to the costs of implementation, lack of reimbursement, questions over liability, lack of real data on cost versus benefit, and how the AI integrates into the workflow and final reports. If radiology departments are looking at several ways to improve efficiency, many are choosing the safe bets of proven technologies. 

"Radiology wants the productivity and efficiency that AI has the potential to give it, but the buyers are increasingly risk averse. They have a limited budget and they can't risk the potential value that an AI algorithm costs compared to a workflow upgrade or a new clinical viewer, something that they know is proven and will benefit value for them as an organization," she explained.

The risk of using AI and not seeing changes in workflow efficiency or patient outcomes, and concerns about liability risk, has caused AI to drop down on many priority lists, she said.  

Reimbursement is a driver for AI adoption in radiology and cardiology

Thompson said very few AI algorithms have earned reimbursement, but the ones that are worthwhile are obvious because they are are widely adopted and already succeeding. 

She said the AI vendor Cleerly and its algorithm to automatically perform a very detailed assessment of soft-plaque burden in coronary arteries is a great example. Thompson said the company is leveraging reimbursement and it now has the potential to shift the paradigm in how cardiology patients may be screened and monitored in the future using low-dose, serial CT scans. 

Cleerly raised $223 million in funding last July to expand its business, which Thompson said is an enormous investment in a startup AI company. "This is one of the largest single amounts of investment received by a vendor in the AI market," she explained. 

In the October 2022 update of the Hospital Outpatient Prospective Payment System, the Centers for Medicare and Medicaid Services assigned a payment rate of $900 to $1,000 for the automated quantification and characterization of coronary atherosclerotic plaque to assess severity of coronary disease using data from coronary computed tomography angiography.

Payment for the new level 3 code, which is the first to provide reimbursement for CT coronary plaque quantification and characterization, became effective Oct. 1, 2022.

Dave Fornell is a digital editor with Cardiovascular Business and Radiology Business magazines. He has been covering healthcare for more than 16 years.

Dave Fornell has covered healthcare for more than 17 years, with a focus in cardiology and radiology. Fornell is a 5-time winner of a Jesse H. Neal Award, the most prestigious editorial honors in the field of specialized journalism. The wins included best technical content, best use of social media and best COVID-19 coverage. Fornell was also a three-time Neal finalist for best range of work by a single author. He produces more than 100 editorial videos each year, most of them interviews with key opinion leaders in medicine. He also writes technical articles, covers key trends, conducts video hospital site visits, and is very involved with social media. E-mail: dfornell@innovatehealthcare.com

Trimed Popup
Trimed Popup