
The Draft Comes First: Radiology Partners, Vision-Language Models and the New Economics of the Read
Dr. Dereck Mush, MD, MBAVision-language models for radiology research.
Cognita Imaging is known for vision-language models for radiology research. It sits in our Diagnostics & Imaging category.
Cognita Imaging is Mosaic Clinical Technologies' advanced AI lab, developing vision-language models for radiology. Mosaic Clinical Technologies, a wholly owned subsidiary of Radiology Partners, reported an internal retrospective study on November 11, 2026 covering 237,905 CT and X-ray exams and 50 conditions. Expert-confirmed enhanced detection averaged 40.9% for CT, 43.4% for chest X-ray and 23.9% for musculoskeletal and abdominal X-ray. Preliminary efficiency improvements exceeded 27% for CT and reached 31% for X-ray. Nearly 8 million investigational drafts had been produced through October 22, 2026. Radiologists review, edit and sign the reports. These are company-reported research findings; the models are investigational and not for sale in the United States. FDA Breakthrough Device designation for Cognita's chest X-ray model is not marketing authorization.
More in Diagnostics & Imaging: Diagnostics & Imaging startups/HealthTech startup database

Dr. Dereck Mush, MD, MBA
Dr. Dereck Mush, MD, MBA
Dr. Dereck Mush, MD, MBA
Dr. Dereck Mush, MD, MBA