svedoliver - Fotolia
RSNA inundated by AI in radiology
At RSNA, AI in radiology was everywhere. Unlike in years past, image analysis was just the tip of the iceberg. Analysts from Frost & Sullivan give their take.
AI in radiology headlined the 2019 Radiological Society of North America's annual meeting where every vendor -- from the biggest players to smaller startups -- had something to offer.
AI in radiology is maturing beyond just image analysis into areas such as R&D tools and intelligent machines, something Siddharth Shah, Frost & Sullivan industry analyst, said he hasn't seen as much of at previous RSNA shows.
"What we've seen this year compared to last year, image analysis is still a major area, but the other areas we have seen ... an increase in terms of applications, especially among the larger companies," Shah said.
One example is how Canon Medical Systems USA Inc., is using AI algorithms. Canon introduced Aquilion ONE / PRISM Edition, a new CT scanner that integrates AI algorithms to boost spectral CT capabilities and automated workflows while providing clinical insights. Philips, another major radiology tech vendor, released an "enterprise imaging portfolio" that included digital imaging platforms, informatics and other services -- all driven by AI.
Still, while the use of AI in radiology is growing, the ability of vendors to demonstrate ROI and efficiency gains is something that's lacking, Shah said.
AI in radiology grows up
Suresh Kuppuswamy, Frost & Sullivan industry analyst for the medical imaging and informatics market, said the major medical imaging equipment vendors talked about more than the hardware they showcased at RSNA. They focused on software and AI applications such as the AI technology Canon has integrated into its new CT scanner, he said.
"Equipment manufacturers were displaying what were the developments on the AI arc that are going to complement their equipment solutions," Kuppuswamy said.
Medical imaging equipment vendors also focused on increasing efficiency with AI in radiology, as well as precision medicine, a model that involves narrowing treatment options for groups of patients based on factors such as genes and socioeconomic status, Kuppuswamy said.
At RSNA, Philips launched its enterprise imaging informatics product that uses AI to enable precision medicine. The product included the IntelliSpace AI Workflow Suite, which can integrate AI applications directly into a clinician's workflow to generate insights from patient data and give providers the ability to "provide a precision diagnosis, leading to targeted therapies with predictable outcomes," Philips said.
Frost & Sullivan industry analyst Srikanth Kompalli echoed Kuppuswamy's observation on efficiency, noting that medical imaging equipment vendors were focused on using AI to boost performance.
For example, GE Healthcare launched more than 30 new imaging intelligent applications aimed at driving clinical efficiency and cost savings, including the Critical Care Suite, a "collection of AI algorithms embedded on a mobile X-ray device for triage," according to a press release.
GE Healthcare said that the Critical Care Suite can improve efficiency by automatically rotating images on portable X-ray machines, while analyzing and flagging certain errors on-device. GE claims the auto rotate AI algorithm can save radiologists at medium to large hospitals more than 70,000 manual clicks annually.
There's more room to grow
Shah said although there continues to be a significant amount of hype around AI in radiology, vendors have a long road ahead when it comes to demonstrating ROI and efficiency gains.
"The challenge is that they don't necessarily focus on what is the clinical use case and how does it really fit into the workflow of a radiologist," he said. "What would be the actual efficiency outcome, efficiency gain, clinical outcome, and how can it really bring in that ROI for the radiologist or hospital administrator? ... If the hospital is paying for it, they need to see the ROI."
More established vendors may have an upper hand here, capable of providing more mature AI models that can produce an ROI and demonstrate potential efficiency gains better than newcomers, Shah said.
Canon for example, a medical imaging equipment company founded in 1948, has a distinct approach to applying AI in radiology, Shah said. Canon is building out AI analysis tools for areas of disease such as stroke, something not all companies are doing, he said.
"What they're saying is, 'We're going to take one disease area, we're going to build out a strategy for that, and then look at the next one," Shah said.