The FDA cleared an AI that color-codes bones on 3D ultrasound, the first step toward spotting fractures without an X-ray

RIVANNA said on Thursday, Oct. 22, 2026, that the FDA has granted 510(k) clearance for BoneEnhance, a machine-learning module for its Accuro XV ultrasound system that segments bone and surrounding tissue and shows skeletal anatomy in 3D during image review.

Most of the world has no X-ray machine within easy reach, and the U.S. government would very much like a way to image broken bones in a disaster zone or a field hospital. That's why BARDA has been paying for this. Today's clearance is the unsexy first rung: the AI labels anatomy, it doesn't call fractures yet. But it's the plumbing that matters, a validated pipeline from robotic scan to AI-labeled image, built so the next module can ride on the same data. If the fracture module clears too, a nurse with an hour of training and a cart could do what now needs a radiology suite. That's a big if, and RIVANNA is careful to say so.

RIVANNA, based in Charlottesville, Virginia, said the FDA has granted 510(k) clearance for BoneEnhance. A 510(k) is the path the agency uses when it decides a device is similar enough to one already allowed on the market. BoneEnhance is a machine-learning module for RIVANNA's Accuro XV diagnostic ultrasound, and RIVANNA says it is the first machine-learning module cleared for that system. It segments bone and surrounding tissue, which means it marks which parts of the scan are bone and which are the tissue around it, and it shows the skeleton in 3D while a clinician reviews the images. Those lines are RIVANNA's.

Accuro XV was cleared in April 2026 for B-mode musculoskeletal and superficial imaging. B-mode is the ordinary grayscale ultrasound picture. The system is portable, with a small footprint. Its probe bends to fit the body and, once the patient is in position, runs a wide 3D scan on its own. The scan does not use ionizing radiation, the kind of energy in an X-ray. Those lines are RIVANNA's.

On each 3D scan, BoneEnhance finds bone and the tissue around it and paints them as color-coded overlays on the 2D ultrasound image. The original image stays visible underneath. A 3D indicator shows which way the scan is oriented and where the slice on screen sits inside the volume. Those lines are RIVANNA's. This clearance is for labeling anatomy. BoneEnhance does not detect fractures, and it does not diagnose an injury.

RIVANNA trained and tested BoneEnhance in its own eight-site U.S. study. Three sites supplied the training data. Five sites supplied an independent test set: 233 volumes, 69,900 images from 173 people. None of that test data was used to train or tune the model. Three board-certified radiologists set the reference labels. Accuracy of the labels, measured per volume, ranged from 95.1% to 100%. Those figures are RIVANNA's. They score how well the software's labels matched the radiologists. They are not a score for finding a fracture.

A separate study, about capturing the images rather than about BoneEnhance's labels, was published in the Journal of Emergency Medicine. In 205 patients, operators with one hour of hands-on training got wrist and ankle images rated adequate for diagnosis in more than 90% of scans, comparable to scans a physician took. Adequate, here, means a reader judged the picture good enough to interpret. That result is about whether a lightly trained person can capture a usable image. It is not a result about finding a fracture.

Next, RIVANNA says, is an investigational injury-detection module meant to help clinicians evaluate possible fractures in an arm or a leg. The study has grown from eight sites to ten to build that data. The module is not cleared. RIVANNA's Oct. 22 announcement does not print a price for BoneEnhance, or a date when hospitals can start using it.

The work is funded in part by BARDA, the biomedical preparedness office inside the U.S. Department of Health and Human Services, under contract 75A50121C00035. Will Mauldin, the chief executive, said, “This clearance is the foundation: the first machine learning layer, anchoring the decision-support capabilities now in development.” That quotation is his.

The picture is RIVANNA's: the Accuro XV system, its review screen showing a wrist scan with color-coded bone labels beside a 3D view of the hand and forearm.

In plain terms, the FDA on Thursday cleared software that color-codes bone on 3D ultrasound from RIVANNA's Accuro XV. It labels anatomy. It does not call a fracture. A fracture-finding module is still in studies and is not cleared. RIVANNA has not published a price or a ship date.