It was no surprise that AI created a buzz at RSNA 2023, but what were the meaningful take-home messages? Dr. Hugh Harvey, managing director of Hardian Health in the U.K., has a list of nine to consider.
- There are very few new startup entrants in the AI space.
- AI vendors have unilaterally downsized their booths, possibly indicating a lack of funding.
- A few new point solutions (clinical decision support CDS or diagnostic) from established vendors but no game changers.
- The talk now is all about "operationalizing" AI and getting it into hospitals fast.
- Mergers and partner maturity are happening at all levels, but it's still early days.
- Everyone is on everyone else's platform. No one knows who is winning.
- There is no robust evidence of return on investment yet. Is this the elephant in the room?
- Large language models (LLMs) are appearing at the fringe, but uncertainty about regulatory processes and skepticism remains high.
- Chicago bars still do great cocktails!
Large multimodal (vision and language) models self-trained on huge datasets of images/reports is another trend identified by Dr. Alexandre Cadrin-Chênevert, a radiologist and computer engineer from Université de Montréal and Saint-Charles-Borromée, Quebec, Canada. "Pairs will eventually replace almost everything in this space," he commented on X.com (formerly known as Twitter).
For more discussion, go to @DrHughHarvey on X.com













![A normal mammogram confirmed by three-year radiologic follow-up illustrates reader-marked regions of interest (ROIs) during (A) unaided (round 1) and (B) artificial intelligence (AI)–assisted (round 2) reading. Each colored dot represents an ROI for recall by a human reader. Readers could mark more than one ROI per case, represented by multiple dots of the same color. During AI-assisted reading, the AI system displayed three visible prompts: two with suspicion of malignancy scores of 35% (left mediolateral oblique [L MLO] and craniocaudal [L CC]) and one with a suspicion of malignancy score of 10% (right craniocaudal [R CC]), shown as polygonal overlays. Without AI, six of 10 readers (60%) marked a false-positive ROI. With AI assistance, this fell to two of 10 (20%). R MLO = right mediolateral oblique.](https://img.auntminnieeurope.com/mindful/smg/workspaces/default/uploads/2026/07/2026-07-14-radiology-mammogram-ai-auto-bias.H0bYO8QlWs.jpg?auto=format%2Ccompress&fit=crop&h=112&q=70&w=112)






