Our prospective, multi-site clinical study with our two university partners has passed its enrollment target. It's a major milestone — and it gives us our first interim look at how Dinmar's AI performs in live clinical workflows on both sides of the Atlantic.
Interim observations
The study remains ongoing and final results will be submitted for peer review. However, the pre-specified interim analysis points in an encouraging direction:
- Faster triage: At sites with AI prioritization active, studies with suspected urgent findings reached a radiologist substantially sooner than under standard first-in, first-out worklists.
- Consistent detection: AI-assisted reads maintained high sensitivity for small findings across both CT and MRI arms, with performance holding up across sites and scanner types.
- Clinician confidence: Participating radiologists reported that heatmaps and automatic measurements made AI findings faster to verify.
What we've learned in the field
Real clinical environments are messier than research datasets — motion artifacts, unusual protocols, incomplete prior studies. Running prospectively has helped us harden our software and improve how we handle edge cases, from automatically detecting low-quality series to gracefully handling missing priors.
"Four years ago this was an idea and a whiteboard. Today our technology is running in hospitals in two countries, helping clinicians see disease sooner."
— Roger Dinmar, Founder & CEOLooking ahead
Next, we'll complete follow-up, finalize the analysis and prepare our results for publication and regulatory discussions. If your health system is interested in participating in future studies, please explore our Early Adopter Program.
Peer-Reviewed: Our First Published Validation Study