Artificial intelligence (AI) technologies had a significant presence at the European Society of Radiology’s annual meeting, the European Congress of Radiology (ECR) 2018. According to a new report published by Signify Research, however, the buzz wasn’t as strong as it was at RSNA 2017 in Chicago.

“AI didn’t seem to generate the same buzz and excitement as it did at RSNA,” wrote author Simon Harris. “Perhaps this is a positive sign that AI is now descending the peak of the hype cycle. Or perhaps it’s a sign that European radiologists aren’t yet fully embracing AI?”

Of course, Harris noted, AI is still clearly gaining momentum. An entire area of the floor at ECR 2018, including a mini theater for presentations, was dedicated to AI.

“There were also a handful of medical imaging AI companies dotted around the main exhibition halls and most of the major vendors found an angle to add AI to their booths,” Harris wrote. “Although there were no major AI-related vendor announcements at ECR, it was evident from walking the exhibition floor that AI continues to make in-roads into medical imaging and the pace of technology commercialization is accelerating.”

Though Harris couldn’t quite put his finger on why there was less hype surrounding AI at ECR 2018, he did note that AI will need to be integrated into radiologists’ existing workflow before it truly becomes mainstream. If vendors developing these state-of-the-art technologies can’t successfully integrate their solutions into PACS, it could end up holding AI back.

“Most generalist radiologists will prefer to access the results from AI algorithms from within their diagnostic viewer, which in most cases today is a PACS,” Harris wrote. “Coming out of PACS to a dedicated AI platform adds an extra step in the process and hence additional time; particularly hard to justify for AI platforms with a narrow offering of algorithms.”

Established use cases, he added, may be an exception to this rule. Breast imagers who are already comfortable using a computer-aided detection workstation are one example, though Harris still believes the industry will continue to see a convergence of such advanced visualization tools and PACS.

Harris also noted that forming partnerships with PACS vendors would be beneficial to providers currently developing AI solutions. “The AI results can be directly overlaid on the images and the radiologist can make edits and annotations,” he wrote. “A tight PACS integration can also give access to the worklist, so that cases can be prioritized in the reading list based on the initial AI findings.”

These are some other key takeaways from the report Harris wrote for Signify Research:

• Imaging biomarkers are becoming more prolific as more quantitative imaging specialists dive into AI. There is still an overall lack of trust from clinicians, however, because they worry about both accuracy and “repeatability.”
• AI companies are starting to think beyond image analysis and consider steps before and after procedures take place. (Before a scan, for instance, AI could help radiologists confirm the patient is placed correctly.)
• Vendors seem focused on making improvements to practice management and workflow for now, with clinical applications viewed as “a longer-term play.”

The entire report from Signify Research, including insight from the firm’s other authors, is available here.