Although RSNA 2023’s theme was “Leading through Change”, there wasn’t much evidence of change in Chicago this year.
Change generally means doing things differently and better, speeding up workflows to increase efficiency. AI companies have been promising this for many years now. But nothing has really progressed yet.
One reason for this is that, in the beginning, algorithms were offered for problems that did not even exist in radiology. This is better now because many new companies involve radiologists who know the real issues.
Secondly, the possibilities for integrating AI into one’s own workflow are anything but easy to understand or implement. Thirdly, the cost models of AI providers are geared more toward the investors’ desire to recoup their money as quickly as possible than to market conditions. This may be different in the US. But in Europe, with its more socially oriented healthcare systems, the absurd costs are a major problem.
Last but not least, hardly any AI provider can offer a broader range of applications. So if you want to use several different algorithms at the same time, this can add up to a considerable amount of monthly costs – especially if you choose the option of integration via a digital platform, it is not even guaranteed that all the desired applications will be available on the same platform.The providers really need to do a lot more work here.
Incidentally, the established modality providers are now taking a relaxed approach. Their AI integrations “directly at the console” are becoming increasingly important. They optimize patient positioning, shorten scan times, optimize the dose and accelerate sequences. The most significant leap forward, however, is being made by ultrasound devices. Whereas you had to have a feel for the transducers in the past, the new systems show you how to hold the probe and in which angulation you can best visualize the organ you are looking for.
Cooperate or die
A trend that can also be seen among established suppliers is that they are entering into partnerships with AI companies or simply buying in expertise, such as Philips and Quibim, Bayer and Blackford, Bracco and Subtle, Agfa and Lunit, Fujifilm and Lunit, Canon and Olea.
But the AI companies are also entering into cooperations or merging. Aidence and Quantib became DeepHealth and are now owned by Radnet, one of the largest radiology companies in the US. The new dogma seems to be “cooperate or die”. It may be that the 2023 AI showcase was bigger than the 2022 showcase, but it was also clear that the size of some companies has shrunk compared to the previous year. This is a clear indication that revenue is still lacking.
Major progress can also be seen in automated, or structured, or guided reporting. What began almost 15 years ago with Word templates has now progressed to Large Language Models and generative AI. It is hard to believe what Philips Speech Recognition has become. In 2011, the former market leader went to Nuance, and Nuance itself was acquired by Microsoft in 2021. Looking back over the last 30 years in radiology, things just keep repeating themselves.
But coming back to the topic of “Leading through Change »…, RSNA 2023 did touch on climate change. With 16 scientific sessions dedicated to sustainability and 11 on climate change that gathered a packed audience, for example, the session ‘What Does Climate Change Have to Do With Radiology?’ featuring radiologist and climate activist Reed Omary, who gave the Opening Lecture at RSNA2022.
But compared with ECR 2023, sustainability was less of a conversation on the technical floor. Maybe there is still some work to be done in that area.
That being said… the RSNA’s annual meeting is always worth the trip. Nowhere else is there so much radiology in such a concentrated form. We will be back at McCormick Place from December 1 to 5, 2024. And we can’t wait to see what has changed between now and next year.