zapp2photo - stock.adobe.com
AI in radiology pushes the move to quantitative imaging, and brings new opportunities, as well as challenges to technology developers from different vendors.
Artificial intelligence has been coming into play in the healthcare industry for a long time, but many still have mixed sentiments toward the possible changes brought by the new technology. Radiology has been no exception.
Getting rid of the overestimation of AI's capabilities in the early stage, radiologists and technology developers are now more realistic about AI's role in enhancing clinical performances by pushing the move to quantitative imaging. Seeing the positive potential of AI in medical imaging, professionals are thinking deeper about how to overcome the challenges of developing and applying AI-powered products. Meanwhile, a new business pattern among vendors is expected to be established in coming years.
In this podcast, Simon Harris, managing director and principal analyst at Signify Research Limited, talks about how AI could provide personalized treatment plans and change the workflow in radiology. He also addresses possible solutions for better interoperability for AI imaging products and predicts the market trend based on his 23 years of experience in technology market intelligence.
Harris points out in the podcast that, "AI addresses many of the fundamental challenges that are facing radiologists and the radiology industry," by improving the efficiency and accuracy of clinical diagnosis using quantitative imaging.
"AI is playing a key part in automatically taking those measurements (when analyzing medical images) and taking them very accurately so there's less variability compared to when a radiologist would do it manually," Harris says.
Consequently, alleviating the shortage of radiologists and saving hospitals' cost on readmissions and follow-up treatments due to previous imprecise diagnosis would be a bonus of applying AI in medical imaging. As to whether those AI-powered tools are affordable for healthcare facilities, Harris says that the best way of thinking about it is not about the price of the software; it's about the return on investment for the healthcare providers.
While people are seeing a lot of potential around AI in medical imaging, challenges such as the lack of interoperability among different AI products and the long process for the FDA to approve those products for sale become the main concerns.
"So typically, today, each algorithm developer has to work quite closely with the PACS vendors to get their products integrated, which is really slowing down the ability of the algorithm developers to bring their products to market and obviously adding extra costs," Harris says.
But he thinks that positive changes are happening in order to solve those problems since the FDA is becoming more educated about AI and vendors are setting up online platforms for integration. One example is EnvoyAI, a medical imaging AI marketplace for radiologists.
From the commercial angle, startups working on AI in medical imaging are dominating the market, while big technology companies like Google are striking into this realm, too. Harris sees it an opportunity for those vendors to cooperate together and set up a new business pattern of partnership, rather than the startups being pushed out of the game.