Healthcare vendors launch AI platforms to secure medical imaging files and automate radiology reporting

Two separate artificial intelligence announcements this week aim to shield medical image files from cyberattacks and ease radiology reporting backlogs. Varist launched a detection engine that scans DICOM files for zero-day malware, while Mosaic Clinical Technologies and DeepHealth each released ambient AI tools to automate radiology workflows as staffing shortfalls widen.

The launches highlight the role of AI for Healthcare in clinical environments, where image-related cyber risks and workforce shortages demand new technical approaches.

Detecting malware in medical image files

The 2025 breach at SimonMed Imaging, where the Medusa ransomware group stole 200GB of data including medical scans affecting more than 1.2 million patients, showed how vulnerable PACS and DICOM files have become, a Varist spokesperson said. The Reykjavik-based vendor built its Hybrid Detection Engine to find malware hidden inside image files that can exploit electronic health record and PACS infrastructure.

The platform handles up to 500 billion file scans per day with a low false-positive rate, according to the company. It uses full file scanning and predictive detection that simulates suspicious behavior, assigning risk scores to flag zero-day exploits not yet catalogued in signature databases. Because the engine runs locally, organizations can avoid uploading protected health information to public clouds, aligning with privacy regulations and cyber insurance requirements.

Siggi Petursson, Varist CTO, said: "A picture is worth a thousand words, especially when lives depend on it, and threat actors may be looking to use that to their advantage. Varist's specialized detection for healthcare environments finds new self-evolving threats designed to evade detection by conventional systems, without adding delays or compromising patients' care and privacy."

Ambient AI targets radiology reporting gaps

The U.S. radiologist shortage is projected to reach about 15% by 2029, even as imaging volumes keep climbing. Mosaic Clinical Technologies introduced MosaicOS + Mosaic Reporting, a cloud-based platform that uses ambient AI to generate structured reports in real time. Radiologists can organize findings, make targeted voice-directed edits, and open cases with preliminary drafts already populated-an area where AI Agents & Automation are shrinking documentation burdens.

"Radiology is facing a chronic shortage of radiologists as imaging volumes continue to climb, leaving radiology practices, departments and the health systems that depend on them with growing backlogs and turnaround time pressure," said Mike Peresie, Mosaic's president. "Mosaic Reporting is designed to help healthcare organizations create reading capacity, improve turnaround times and better meet service levels that referring physicians and patients expect."

Separately, DeepHealth, RadNet's informatics division, announced that its AI-powered reporting platform, Reporting Pro, is now commercially available and being deployed at scale. The system pulls clinical findings from FDA-cleared third-party image diagnostic tools directly into draft structured reports. Dr. Jason Sinner, a radiologist and medical director at RadNet, said having findings already populated "significantly reduces reporting times, translating directly to faster turnaround times for patients and referring physicians who rely on receiving reports on a timely basis in order to make critical treatment decisions."

Why this matters for healthcare professionals

For imaging directors, PACS administrators, and radiologists, these tools address two simultaneous pressures: locking down medical image files against ransomware groups that now use them for double-extortion, and cutting report turnaround when there aren't enough radiologists to handle caseloads. Local scanning satisfies compliance mandates, while ambient reporting embedded directly into workflows can save minutes per case. The results are moving fast-DeepHealth expects its first RadNet customers to go live within the next quarter, and Varist's engine already scans at hyperscale.