AI Analysis of Routine CT Scans Could Help Detect Weak Bones Earlier Than Ever

Advances in medical imaging have opened a surprising new door: routine CT scans, originally taken to check for problems like kidney stones, lung nodules, or abdominal pain, can now be repurposed to detect early signs of bone loss. A large research project from NYU Langone Health, developed with experts at Visage Imaging, shows how artificial intelligence can transform existing scans into powerful tools for identifying weakened bones long before a fracture occurs. This simple shift in how hospitals analyze their imaging data could dramatically change the landscape of osteoporosis detection.

The study, published in the journal Radiology, looked at an enormous dataset: 538,946 CT scans from 283,499 patients. These scans came from 43 different CT machine models and followed standard imaging protocols, making the research reflect real-world medical practice rather than lab-perfect conditions. The AI tool was trained to analyze vertebrae throughout the lumbar and thoracic spine, measuring bone mineral density and comparing it against values adjusted for age, sex, and race or ethnicity. Radiologists then checked the AI-generated measurements to confirm their accuracy.

One of the most interesting aspects of the findings is how clearly the data highlighted bone-density differences across demographic groups. Contrary to common assumptions, women under 50 actually had higher bone density than men in the same age group. However, after age 50, this pattern reversed. Women — especially postmenopausal women — experienced a steeper decline, leaving older men with higher bone density overall. The results also showed racial and ethnic differences: Black patients had the highest bone density, followed by Asian patients, with White patients showing the lowest average values in the dataset.

This kind of population-level insight is usually only available after dedicated bone-scanning procedures like DEXA scans, which are not performed nearly as often as guidelines recommend. Osteoporosis is estimated to affect more than 10 million Americans, with another 40 million showing early signs of low bone mass. Many people do not know they have weakened bones until they suffer a fracture — and fractures, especially in older adults, can be life-altering or even life-threatening. That’s why the potential of opportunistic CT screening is so significant.

By using scans that patients are already getting for unrelated medical issues, AI can flag individuals who may have low bone density or early osteoporosis. This gives healthcare providers a chance to intervene sooner with lifestyle changes, supplements, medications, or further testing. The researchers argue that using the “big data” already sitting in hospital imaging databases could meaningfully reduce the enormous rate of underdiagnosis in osteoporosis.

This isn’t the first time that AI-assisted “opportunistic screening” has shown big promise. Earlier work from the same team demonstrated that using existing abdominal CT scans for bone-density evaluation could more than double the number of patients assessed each year. That earlier study also projected potential savings of over $2.5 billion annually in Medicare costs by preventing fractures through earlier detection and treatment. The new work builds on that momentum by validating the approach across a wider variety of CT scanners and patient backgrounds, demonstrating that the method can work at scale.

The research team emphasizes that AI isn’t replacing radiologists — it’s simply doing the heavy lifting behind the scenes. The algorithm rapidly examines thousands of vertebrae, measures density, and spots subtle patterns. Radiologists then step in to verify and interpret the findings. This partnership makes it possible to screen enormous patient populations automatically without burdening hospital staff.

This new AI tool is expected to be used soon in NYU Langone’s clinical trial program, allowing hospitals to offer opportunistic osteoporosis screening as part of routine care. If widely adopted, it could fundamentally change how osteoporosis is diagnosed and treated.

Beyond bone health, the research team is already looking ahead. They plan to use similar datasets to train AI models capable of identifying cardiovascular disease, blood vessel abnormalities, and muscle loss — all conditions that traditionally require separate tests or dedicated imaging. Past investigations by the group found that abdominal CT scans could reveal hidden cardiovascular risks, meaning the same image could someday be read for multiple purposes by AI.

This idea — using a single medical scan to detect several unrelated health problems — represents a significant shift in how imaging may work in the future. With AI, every scan becomes more valuable, and patient care becomes more preventive rather than reactive.

Understanding Bone Density and Osteoporosis: A Helpful Breakdown

To give readers more context, here’s a straightforward explanation of what bone density means and why it matters.

Bone Mineral Density (BMD):
This is a measurement of how much mineral is packed into a section of bone. Higher density means stronger bones. Lower density signals conditions like osteopenia or osteoporosis, which raise the risk of fractures.

How CT Scans Detect Bone Density:
A CT scan creates detailed cross-section images of the body using X-rays. Bones appear clearly, and AI can measure their internal density by analyzing pixel values. Although CT was not originally created for bone-density screening, AI allows precise measurements similar to what you’d get from dedicated scans.

Why Women Lose Bone Faster After 50:
Estrogen plays a major role in maintaining bone strength. After menopause, estrogen levels drop sharply, causing accelerated bone loss — which is why osteoporosis remains much more common in older women.

Racial Differences in Bone Density:
Research consistently shows that bone density tends to be highest in Black populations. Genetics, bone structure, and lifestyle factors all play a role in these differences.

Why Opportunistic Screening Matters:
Many people do not get screened for osteoporosis until after a fracture. Opportunistic screening solves this problem simply by re-reading scans that already exist, requiring no new radiation exposure or new appointments.

The Future of Opportunistic Imaging

The most exciting aspect of this development is not just what it means for bone health — but what it demonstrates about the future of medical imaging. If AI can analyze a CT scan for bone density and also detect signs of cardiovascular disease or muscle loss, then hospitals can dramatically improve early detection using the scans they are already generating every day.

This could help catch silent conditions early, reduce healthcare costs, and allow patients to take action before symptoms appear. It represents a shift from reactive medicine to proactive, data-driven healthcare.

The NYU Langone team’s work is among the largest studies ever conducted on AI-assisted CT bone analysis, and its findings provide a strong foundation for expanding this idea across medicine. With hundreds of thousands of scans analyzed, the data strongly suggests that opportunistic CT analysis is not just feasible — it may soon be essential.

Research Paper:
Deep Learning-Based Opportunistic CT Osteoporosis Screening and Establishment of Normative Values
https://doi.org/10.1148/radiol.250917