← The Newsroom
Dr. Carpenter's Take · The Bone Health Brief

The Scan You Already Had Might Save Your Hip

Deep learning can read bone density from CT scans patients already had, turning routine imaging into early warning.

SC
Dr. Shannon CarpenterFounder & CEO · November 15, 2025

This may be my favorite kind of progress: not a flashy new machine, but a smarter way to use the ones we already have. A November 2025 study trained a deep-learning method to read bone density from routine CT scans, pulling signal from roughly 539,000 exams people had already undergone for entirely different reasons, and using them to flag low bone mass and set reference values for what's normal.

Think about how many CT scans happen every day for chest pain, abdominal complaints, injuries, screening. Each one already contains a quiet picture of the patient's bones. This approach lets us read that picture instead of ignoring it, “opportunistic” screening, catching low bone density in people who never came in asking about it.

Here's how I describe it to patients: a tool like this is a smoke detector, not the fire department. It can warn you the risk is there. It cannot, by itself, prevent the fracture.

The detector earns its keep only when the alarm leads to action, a real evaluation and a real plan. So if a scan ever surfaces a concern about your bones, treat it as a gift of early warning, and follow it with a conversation.

Early knowledge is only powerful when we use it. That's the whole idea behind “Demand the Scan”: don't wait for the break to take bone health seriously.

Wondering about your own bone health?

Book a Consultation
The research behind this

An automated deep-learning method flags low bone mineral density on routine CT scans, drawing on roughly 539,000 exams to set diagnostic thresholds and normative values, pointing toward “opportunistic” screening from imaging patients already receive.

Westerhoff M, Gyftopoulos S, Dane B, et al. Deep Learning–based Opportunistic CT Osteoporosis Screening and the Establishment of Normative Values. Radiology. 2025;317(2). doi:10.1148/radiol.250917
Read the source study ↗