- Campbell Arnold
- Sep 2
- 3 min read
Updated: Sep 9
“In healthcare, adoption happens at the speed of trust.”
— Kevin Field, CRO Rad AI
Welcome to RadAccess: Impressions, your quick-read companion to the main RadAccess newsletter. Here we only deliver the essential information. For more details, you can always turn to the full RadAccess newsletter.
In this issue, we cover:
Dialing in Image Generation: Metadata-Guidance during Synthesis
Harrison.ai Reveals Preliminary Foundation Model Trial Results
Halve Your Scan Time, Same Great Quality
Rad AI Launches New Blog, the Readout
If you want to stay up-to-date with the latest in Radiology and AI, then don't forget to subscribe!
Dialing in Image Generation: Metadata-Guidance during Synthesis
A recent Cell Reports Medicine article introduced a new metadata-guided image synthesis framework called TUMSyn, which brings a new level of control to MRI generation. Instead of ignoring scan variability, the method embeds metadata like acquisition parameters and patient demographics directly into the synthesis process, producing high-fidelity images that can be tailored by voxel size, echo time, or even patient age. The algorithm was developed and validated using over 30K brain scans across 13 datasets, with performance that matched or exceeded other leading algorithm, all while preserving alignment with conditioned metadata. By enabling parameter-specific customizable synthetic images, this approach could transform both clinical deployment and algorithm training. Best of all, the code is available on GitHub.
Harrison.ai Reveals Preliminary Foundation Model Trial Results
Harrison.ai’s foundation model Harrison.rad.1 is being tested in the US Healthcare AI Challenge, where AI-generated chest X-ray reports are directly compared to radiologist-written ones. In preliminary results from nearly 3,000 evaluations of 117 general chest X-ray exams, the model achieved a 65% clinical acceptability rate versus ~80% for human-authored reports. While this findings demonstrates that AI-generated reports aren't ready to replace radiologists, it also shows real-world promise for using foundation models to draft reports and streamline radiology workflows. Honestly, it's immensely impressive that 2/3rd of AI-generated reports for this broad task were deemed clinically acceptable by radiologists!
Halve Your Scan Time, Same Great Quality
A new Radiology Advances study shows that Siemens’ Deep Resolve DL-accelerated MRI exams can cut scan times in half (54s vs 111s) while maintaining or even slightly improving image quality. Although some unique artifacts were noted, they didn’t impact diagnostic accuracy. Faster, high-quality scans could help reduce wait times, ease scheduling bottlenecks, and improve the patient experience.
Resource Highlight: The Readout by Rad AI
Rad AI, one of the largest radiology generative AI companies serving nearly half of U.S. health systems, has launched The Readout, a new blog offering deeper insights into imaging and AI. Instead of product updates, it features long-form perspectives on clinical workflows, industry trends, and voices from across the company, from researchers to executives. definitely worth checking out a few posts.
Feedback
We’re eager to hear your thoughts as we continue to refine and improve RadAccess. Is there an article you expected to see but didn’t? Have suggestions for making the newsletter even better? Let us know! Reach out via email, LinkedIn, or X—we’d love to hear from you.
References
Wang, Yulin, et al. "Toward general text-guided multimodal brain MRI synthesis for diagnosis and medical image analysis." Cell Reports Medicine (2025).
Lyo, Shawn K., et al. "Deep Learning MRI Halves Scan Time and Preserves Image Quality Across Routine Neuroradiologic Examinations." Radiology Advances (2025): umaf029.
Disclaimer: There are no paid sponsors of this content. The opinions expressed are solely those of the newsletter authors, and do not necessarily reflect those of referenced works or companies.