{"id":19235,"date":"2026-09-07T05:18:19","date_gmt":"2026-09-07T08:18:19","guid":{"rendered":"https:\/\/rtmedical.com.br\/tmp-en-1788769098746\/"},"modified":"2026-09-07T05:18:26","modified_gmt":"2026-09-07T08:18:26","slug":"vara-autonomous-triage-mammography","status":"publish","type":"post","link":"https:\/\/rtmedical.com.br\/en\/vara-autonomous-triage-mammography\/","title":{"rendered":"Vara Wins CE Mark for Autonomous Mammography Triage"},"content":{"rendered":"

Vara<\/strong>, a Berlin-based AI company, received CE certification on September 2, 2026 under the EU Medical Device Regulation (EU MDR<\/strong>, Class IIb) for autonomous triage<\/strong> in organized breast cancer screening programs. It is the first breast AI anywhere in the world certified to report screening mammograms as normal without a radiologist reading them. The rule is simple: exams the algorithm classifies as clearly normal are reported by the AI itself; every other exam goes to at least one radiologist.<\/p>\n

\"Breast
Vara’s CE certification lets the AI report clearly normal screening exams on its own. Image: Radiology Business<\/figcaption><\/figure>\n

The news was reported by Radiology Business and confirmed in a company press release. The certification covers 2D digital mammography and 3D tomosynthesis, and the system analyzes up to three rounds of a woman’s prior exams. Vara says autonomous triage is now available to screening programs across Europe, with rollout in each country following national guidelines and the clinical readiness of each service.<\/p>\n

What Autonomous Triage Changes in Mammography Screening<\/h2>\n

Organized European programs rely on double reading<\/strong>: two radiologists independently assess each mammogram, with arbitration when they disagree. The model raises sensitivity but doubles reading volume at a time when a shrinking radiologist workforce is straining programs. Autonomous triage removes the human read only from the fraction of exams the AI flags as clearly normal; for those cases, it replaces one or both reads.<\/p>\n

The step is bigger than it looks. In October 2025, Vara had received a CE mark as an independent second reader<\/strong>: the AI took the seat of one of the two radiologists, but a human still read every exam. Now, a share of exams leaves the workflow without any physician involvement. No regulator, including the FDA, had authorized anything like it in breast imaging.<\/p>\n

ATMON: The System That Watches the AI in Real Time<\/h2>\n

The centerpiece of the certification is ATMON<\/strong> (Autonomous-Triage Monitoring), a continuous supervision system that sets and tracks each site’s operating point. It monitors changes in mammography hardware, system health and daily performance signals. When any indicator moves outside defined limits, the site automatically reverts to 100% radiologist reading.<\/p>\n

Vara frames ATMON as the way to meet the continuous human oversight the EU AI Act<\/strong> requires of high-risk AI systems, a category that includes medical devices regulated under the MDR. Stefan Bunk, CTO and co-founder, said the system “is based on seven years of continuous monitoring of how our AI performs in the real world, on every single case.”<\/p>\n

Jonas Muff, CEO and co-founder, summed up the reasoning: “Studies have shown that AI can improve cancer detection. But even the best AI model’s performance can shift once it operates outside the controlled conditions of a study. For AI to deliver its positive impact on patient outcomes, we need to notice when this happens. We believe our Atmon system can be a template for deploying AI safely across healthcare.” ATMON will also be offered to existing customers that do not adopt autonomous triage.<\/p>\n

PRAIM: The Evidence Base With 461,000 Women<\/h2>\n

The certification rests on the PRAIM<\/strong> study, published in Nature Medicine in January 2025 with Nora Eisemann as first author. It was a prospective, multicenter, observational non-inferiority implementation study run at 12 sites of the German program between July 2021 and February 2023, enrolling 461,818 women aged 50 to 69 and 119 radiologists, with no exclusion criteria. It remains the largest prospective study of AI in mammography screening published to date.<\/p>\n

In the AI-supported group, the cancer detection rate<\/strong> was 6.7 per 1,000 women versus 5.7 without AI, a relative increase of 17.6% (95% CI: 5.7% to 30.8%) that Radiology Business rounded to “nearly 20%”. Recall rates did not rise (37.4 vs. 38.3 per 1,000), and the positive predictive value of recall improved from 14.9% to 17.9%. One detail matters: the AI tagged 56.7% of exams as normal, but in PRAIM radiologists still reviewed those cases. Autonomous triage is precisely the step the study did not take.<\/p>\n

PRAIM was funded by Vara, and three authors are company employees. The conflict is disclosed in the paper, but it belongs in any reading of the results.<\/p>\n

How It Compares With MASAI and Other Trials<\/h2>\n

The Swedish randomized MASAI<\/strong> trial, with more than 105,000 women, took a different route: AI routes low-risk exams to single reading and high-risk exams to double reading. Results in The Lancet Digital Health in 2025 showed 29% more cancers detected and a 44% lower screen-reading workload; the final publication in The Lancet in January 2026 reported 1.55 interval cancers per 1,000 women versus 1.76 in the control arm. The authors, led by Kristina L\u00e5ng, were explicit that at least one radiologist stays on every exam.<\/p>\n

We have covered studies suggesting that AI could cut 77% of mammograms from double reading<\/a> and that mammography AI cuts radiologist workload by 64%<\/a>. We have also reported that many radiologists feel breast AI falls short of their expectations<\/a>. Vara goes further than all of them by removing the human from a share of exams, which is exactly why it needs a safety barrier like ATMON.<\/p>\n

What It Means Outside Germany<\/h2>\n

The UK<\/strong>, the Nordic countries, the Netherlands and Spain all run double-reading programs facing the same workforce squeeze, and several are already testing AI as a second reader. For them, autonomous triage is the natural next question, one that national guideline committees will now have to answer. In the United States<\/strong>, screening mammograms are read once, the FDA has never cleared autonomous AI for mammography, and groups like Radiology Partners are still asking the FDA to clarify imaging AI rules<\/a>. Vara says it is not yet available there and remains focused on the EU.<\/p>\n

The legal question is universal: who is responsible for an interval cancer in an exam reported by software alone? In most jurisdictions, the radiology report is a medical act tied to a licensed physician. Certification under the MDR settles device safety, not liability, and that conversation is only beginning.<\/p>\n

Outlook and Limitations<\/h2>\n

Vara says more than 50% of Germany’s organized screening programs run on its platform, processing over 250,000 exams a month. Founded in 2018, the company has raised roughly $34 million, including a round of nearly $9 million in 2024 that funded its expansion into India.<\/p>\n

The essential question remains open: there is no peer-reviewed publication yet evaluating autonomous triage in live operation. The certification rests on PRAIM, which kept radiologists on every exam, and on the company’s own accumulated monitoring data. Performance in populations whose cancer prevalence, breast density and equipment differ from Germany’s is also unknown. ATMON promises to catch drift; whether it catches it in time is something only years of use will show. For a broader view of where the field stands, see our roundup of the breast imaging studies that defined 2026<\/a>.<\/p>\n

Source:<\/strong> Radiology Business<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

Vara earns CE certification (MDR Class IIb) for AI to report normal screening mammograms without a radiologist. See how ATMON works and what comes next.<\/p>\n","protected":false},"author":1,"featured_media":19222,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"om_disable_all_campaigns":false,"_monsterinsights_skip_tracking":false,"_monsterinsights_sitenote_active":false,"_monsterinsights_sitenote_note":"","_monsterinsights_sitenote_category":0,"ngg_post_thumbnail":0,"_rt_cluster":"","fifu_image_url":"","fifu_image_alt":"","footnotes":""},"categories":[102,100],"tags":[],"class_list":["post-19235","post","type-post","status-publish","format-standard","has-post-thumbnail","category-ai","category-radiology"],"aioseo_notices":[],"rt_seo":{"title":"","description":"Vara earns CE certification (MDR Class IIb) for AI to report normal screening mammograms without a radiologist. See how ATMON works and what comes next.","canonical":"","og_image":"","robots":"index,follow","schema_type":"Article","include_in_llms":true,"llms_label":"Vara: CE certification for autonomous mammography triage","llms_summary":"Berlin-based Vara received CE certification on Sept 2, 2026 under EU MDR Class IIb for autonomous triage in organized breast screening: clearly normal exams are reported by AI without a radiologist. The ATMON system monitors performance in real time and reverts sites to full radiologist reading on drift. Evidence base: PRAIM (Nature Medicine 2025, 461,818 women, +17.6% cancer detection). Not available in the US.","faq_items":[],"video":[],"gtin":"","mpn":"","brand":"","aggregate_rating":[]},"_links":{"self":[{"href":"https:\/\/rtmedical.com.br\/en\/wp-json\/wp\/v2\/posts\/19235\/"}],"collection":[{"href":"https:\/\/rtmedical.com.br\/en\/wp-json\/wp\/v2\/posts\/"}],"about":[{"href":"https:\/\/rtmedical.com.br\/en\/wp-json\/wp\/v2\/types\/post\/"}],"author":[{"embeddable":true,"href":"https:\/\/rtmedical.com.br\/en\/wp-json\/wp\/v2\/users\/1\/"}],"replies":[{"embeddable":true,"href":"https:\/\/rtmedical.com.br\/en\/wp-json\/wp\/v2\/comments\/?post=19235"}],"version-history":[{"count":1,"href":"https:\/\/rtmedical.com.br\/en\/wp-json\/wp\/v2\/posts\/19235\/revisions\/"}],"predecessor-version":[{"id":19237,"href":"https:\/\/rtmedical.com.br\/en\/wp-json\/wp\/v2\/posts\/19235\/revisions\/19237\/"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/rtmedical.com.br\/en\/wp-json\/wp\/v2\/media\/19222\/"}],"wp:attachment":[{"href":"https:\/\/rtmedical.com.br\/en\/wp-json\/wp\/v2\/media\/?parent=19235"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/rtmedical.com.br\/en\/wp-json\/wp\/v2\/categories\/?post=19235"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/rtmedical.com.br\/en\/wp-json\/wp\/v2\/tags\/?post=19235"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}