The Food and Drug Administration has granted 510(k) clearance to DeepHealth Breast Ultrasound, an AI-powered system built by RadNet’s DeepHealth division that automates lesion detection, characterization, and reporting during one of the most common but most operator-dependent breast imaging exams. The clearance, announced July 30, 2026, positions the tool as part of a broader push to standardize breast ultrasound, a scan roughly 40% of women undergo at some point in their lives, often as a supplemental check when mammography alone is not enough.
Why Ultrasound Needed an AI Fix
Breast ultrasound has long been considered essential to the breast care pathway but notoriously inconsistent in practice. Dr. Jason McKellop, medical director of women’s imaging for RadNet California, described it as “a highly complex, operator-dependent examination, which can lead to significant variability in image acquisition, interpretation and reporting.” Unlike mammography, where the imaging technique is fairly standardized, ultrasound results can vary depending on who is holding the probe and how a lesion is measured and described, making consistent follow-up recommendations harder across large imaging networks.
The Clinical Validation Numbers
DeepHealth’s FDA submission included a multi-reader, multi-case clinical validation study involving 16 U.S. board-certified radiologists. The results showed greater than 98% accuracy in localizing breast lesions, an 8 percentage-point improvement in cancer detection sensitivity, and a 37% reduction in radiologist interpretation time when the AI tool assisted the read. For imaging networks handling thousands of ultrasound studies a month, that interpretation-time drop translates directly into radiologist capacity that can be redirected toward other exams or faster turnaround on results for patients.
How the System Fits Into a Broader Breast Suite
The ultrasound clearance builds on DeepHealth’s existing Breast Suite platform, which already combines cancer detection with risk stratification tools designed to flag which patients may need additional screening beyond standard mammography. Kees Wesdorp, DeepHealth’s CEO, called the ultrasound clearance “a pivotal step toward a new, AI-powered standard of care in breast cancer screening and diagnostic pathways,” arguing that pairing detection algorithms with workflow tools lets radiologists “detect cancers earlier, with more confidence.” DeepHealth operates as part of RadNet, one of the largest outpatient imaging networks in the United States, giving the new tool an immediate path to deployment across hundreds of existing imaging centers rather than requiring a slow hospital-by-hospital sales process.
Part of a Broader Wave of AI Cancer-Detection Clearances
The breast ultrasound clearance arrives amid a busy stretch for FDA-cleared AI cancer tools in 2026. Artera received clearance in May for ArteraAI Breast, aimed at early-stage, hormone receptor-positive, HER2-negative invasive breast cancer, while Ibex Prostate Detect won 510(k) clearance as a digital pathology tool that helps identify small or rare prostate cancers in biopsy tissue. Separately, the FDA has cleared AI software specifically credited with boosting cancer detection in dense breast tissue, a population long known to be harder to screen accurately with mammography alone because dense tissue can mask tumors on standard X-ray images.
Two Views on What Standardized AI Reads Mean for Patients
Supporters, including radiologists quoted in trade coverage from AuntMinnie and Radiology Business, frame the clearance as a direct patient-safety win: fewer missed lesions and less variability between sonographers means fewer women falling through the cracks of an exam known for operator dependence. Critics of rapid AI imaging rollouts caution that clinical validation studies, even multi-reader ones with 16 radiologists, are still smaller and shorter than the real-world populations these tools will eventually screen, and that sensitivity gains measured in controlled study conditions do not always hold up once a tool is deployed across diverse patient populations, breast densities, and ultrasound machine models nationwide.
What Comes Next
RadNet has signaled plans to roll the cleared tool out across its own network of imaging centers first, giving the company a built-in test bed to track real-world performance before wider licensing to outside hospital systems. The clearance also adds pressure on competing AI imaging vendors targeting breast cancer detection, prostate pathology, and dense-tissue screening to bring their own FDA submissions forward, as radiology practices increasingly treat AI-assisted reads as a baseline expectation rather than an experimental add-on. For patients, the more immediate effect may simply be faster turnaround on ultrasound results and more consistent follow-up recommendations regardless of which sonographer performed the scan.
