SwRI develops tool to predict osteoporotic fracture risk

SAN ANTONIO — February 11, 2025 — Southwest Research Institute (SwRI) has developed new technology to enhance clinical bone imaging and decrease osteoporotic fractures in the elderly. The tool, described in a new SwRI-led study, uses artificial intelligence (AI) to create super-resolution images showing the inner structures of bones in great detail to better determine risk for fracture.

SwRI found that in almost all cases, the super-resolution (SR)-enhanced images were more accurate than X-rays or CT scans in quantifying bone characteristics that are indicative of strength and fracture risk.

Osteoporotic fracture in the elderly is a widespread problem. Half of women over 55 suffer some sort of fracture at some point. Of the millions of people who break a hip each year, 50% never walk again on their own, and 20% die within a year.

“Engineers and orthopedic clinicians are trying to identify people at risk because there are interventions that can make bones stronger, but it's all about identifying who is at risk,” said SwRI’s Dr. Lance Frazer, the study’s lead author. “Currently, the most common method to diagnose fracture risk is an X-ray or a CT scan, but the resolution of both imaging techniques is too low to really determine the structural strength of the bone.”

For this study, SwRI used a type of AI known as super-resolution technology, a set of techniques used to make low-quality images sharper and more detailed. SwRI’s deep-learning SR approach trains AI to map complex patterns and textures by teaching it to observe details in several examples. As a result, it can use low-resolution image data of bones to predict what a much higher resolution image of that bone would look like.

“Once we have that high-resolution image data, we have many tools at our disposal to predict just how strong the bone actually is,” Frazer said. “We can see if the bones are susceptible to fracture and if it’s necessary to intervene to prevent that. After all, bones are adaptable. Once a fracture risk is identified, a doctor can recommend medication or exercises to strengthen vulnerable bones and keep an injury from occurring.”

Frazer suggested applying SR technology to medical imaging equipment, which would provide a clinician raw medical images, such as X-rays, as well as processed images enhanced by AI. The resulting images are sharper and clearer, potentially accompanied by additional diagnostics to help physicians make better medical decisions.

“Everybody falls. Everyone over the age of 55, and particularly those over 65, will fall at some point, so understanding bone strength for this population is important,” Frazer said. “A lot can be done in terms of fall prevention. However, the imminent question remains: when you fall, are you at risk for a fracture? Knowing this is critical.”

The paper “Super-resolution of clinical CT: Revealing microarchitecture in whole bone clinical CT image data” appears in the journal Bone and can be accessed at https://www.sciencedirect.com/science/article/abs/pii/S8756328224001042.

For more information, visit https://www.swri.org/markets/biomedical-health/biomedical-devices/biomechanics-human-performance/bone-biomechanics.