Cognita Imaging

Companion research papers from Mosaic Clinical Technologies and Cognita on investigational vision-language models for CT and X-ray report drafting, evaluated on 237,905 exams across 50 conditions.

Radiology sits at the front line of diagnosis, is facing a growing global workforce shortage, and is already deeply digitized. Together, these realities make it uniquely positioned for AI to drive meaningful, large-scale change.

Around the world, millions wait weeks or months for a diagnosis because specialists are in short supply especially in radiology, the diagnostic core of medicine.

Cognita develops AI systems to increase the world's access to healthcare, starting with radiology.

Mosaic Clinical Technologies’ vision-language models (VLMs), developed by its AI lab Cognita Imaging, detect additional clinically important findings in over 237,000 CT and X-ray exams. The models, tested across 50 conditions, improve detection by identifying findings confirmed by expert radiologists. The study evaluates efficiency and accuracy gains in radiology workflows using AI-powered analysis.

Nov 2026thecanadianpressnews.caNews article

Cognita Imaging secures an FDA contract to assess large language models (LLMs) in evaluating AI-generated radiology reports. The project supports the FDA’s exploration of regulatory frameworks for generative AI-enabled medical devices. The focus lies on validating AI-driven report accuracy and safety in radiology applications.

Sep 2026MedTech DiveNews article

Cognita Imaging secures a **$1.29 million FDA contract** to assess a novel method for evaluating generative AI in radiology report generation. The project aims to refine how AI outputs are validated for clinical use. Cognita, a subsidiary of Mosaic Clinical Technologies, will collaborate with the FDA to develop and test these evaluation frameworks.

Sep 2026Radiology PartnersNews article

Blankemeier and co-founders sold their healthcare AI startup after one year, having developed models that generate radiology reports from medical images like X-rays and CT scans. Their system interprets tens of thousands of potential diagnoses, mimicking radiologists’ clinical reasoning—a broader approach than existing AI tools limited to detecting specific conditions. The sale accelerated real-world deployment of their Ph.D.-derived research.

Apr 2026Crunchbase NewsNews article