Vara Wins First EU Approval for AI Mammogram Triage Without Radiologist Review

A Berlin-based AI company has secured what it claims is the world's first regulatory approval for a breast cancer screening tool that can classify certain mammograms as normal without any radiologist review.

Vara announced September 2 that it received CE Certification under the European Union's Medical Device Regulation for its autonomous triage product. The approval allows the system to independently read and clear normal mammograms in organized population screening programs, while all other exams continue to require interpretation by at least one radiologist.

The certification addresses a critical challenge in European breast screening, where programs commonly require two radiologists to read every mammogram—a practice that strains an already shrinking workforce.

How the system works

Vara's autonomous triage operates within a broader platform called Atmon (Autonomous-Triage Monitoring), which continuously tracks the AI's real-world performance on every case. The monitoring system sets and adjusts each imaging site's operating parameters in real time while tracking changes in mammography hardware, system health, and daily performance metrics.

If performance signals move outside defined limits, Atmon can automatically revert a site to full radiologist reading. This continuous oversight mechanism aligns with requirements for high-risk AI systems under the EU AI Act, according to Vara.

"What made this possible is not the AI model alone," said Stefan Bunk, Vara's chief technology officer and co-founder. "It is based on seven years of continuous monitoring of how our AI performs in the real world, on every single case."

Clinical evidence and deployment

Vara's certification draws on real-world performance data, including a Nature Medicine study published last year. That research—which the company describes as the largest prospective study of AI in breast cancer screening to date—found Vara's tools increased breast cancer detection by nearly 20% across approximately 462,000 women at a dozen German sites.

"Even the best AI model's performance can shift once it operates outside the controlled conditions of a study," noted Jonas Muff, Vara's CEO and co-founder. "For AI to deliver its positive impact on patient outcomes, we need to notice when this happens."

Founded in 2018, Vara estimates that over 50% of Germany's organized breast screening programs currently use its products, processing more than 250,000 screening exams monthly. The company's technology handles both 2D mammography and 3D tomosynthesis and can analyze up to three rounds of prior exams.

Why it matters

This approval represents a significant shift in how AI can be deployed in clinical radiology. Rather than serving as a second reader or decision-support tool, Vara's system can make autonomous diagnostic decisions for a subset of cases. The move directly addresses Europe's radiologist workforce shortage while establishing a framework for continuous AI monitoring that could inform future regulatory approaches. However, the autonomous model also raises questions about liability, quality assurance, and the appropriate balance between efficiency gains and human oversight in medical imaging.

Availability and expansion

Autonomous triage is now available to screening programs across Europe, with rollout dependent on each country's national guidelines and clinical readiness. Vara plans to make its Atmon monitoring system available to existing customers who may not adopt autonomous triage, allowing them to benefit from the safety infrastructure.

The company has raised approximately $34 million in financing, including nearly $9 million in 2024 when it expanded into India. Vara's technology is not yet available in the United States, as the company remains focused on the European market.

These details were first reported by Radiology Business.