Automated bone age estimation from routine abdominal CT using AI

Development and Evaluation of a Deep Learning-Based Model for Automated Osteoporosis Assessment Using CT Images

This project will try to use routine noncontrast abdominal CT scans and deep learning to estimate bone age in adults who had these scans for other medical reasons.

Quick facts

• Study type: Observational
• Enrollment: 3000 (estimated)
• Ages: 18 Years and up
• Sex: All
• Sponsor: Peking University People's Hospital (Academic / other)
• Drugs / interventions: radiation
• Locations: 1 site (Beijing)
• Trial ID: NCT07162168 on ClinicalTrials.gov

What this trial studies

This retrospective analysis uses existing noncontrast abdominal CT scans that include the proximal femur to train and test a deep-learning model to estimate bone age and signs of osteoporosis. Investigators will include adult scans taken for non-orthopedic indications and exclude images with implants, tumors, severe artifacts, or post-surgical changes. Demographic data such as chronological age and sex will be used alongside image features to compare model output to known patient information and any available bone-health measures. By repurposing routine imaging, the approach aims to provide bone-health information without additional radiation or dedicated scans.

Who should consider this trial

Good fit: Ideal candidates are adults over 18 whose routine noncontrast abdominal CT scans fully include the proximal femur, were performed for non-orthopedic reasons, and have available demographic data, with no hip implants, tumors, severe deformity, or prior femoral fractures.

Not a fit: Patients without suitable CT images (for example, scans that do not include the proximal femur, have severe artifacts, or who have hip implants, bone tumors, major deformity, or prior femoral surgery) and pediatric or pregnant patients would not be expected to benefit from this analysis.

Why it matters

Potential benefit: If successful, this approach could provide patients and clinicians with bone-age and osteoporosis risk information from existing CTs without extra scans or radiation, supporting earlier and more personalized bone-health care.

How similar studies have performed: Previous work using opportunistic CT and deep learning to detect osteoporosis and estimate bone density has shown promising results, though direct bone-age estimation from routine noncontrast abdominal CT is a relatively novel application.

Eligibility criteria

Inclusion Criteria:

* Adults aged over 18 years.
* Underwent routine noncontrast abdominal CT scans.
* CT scans fully included the proximal femur.
* Scans were performed for non-orthopedic clinical indications.
* Provided necessary demographic information (e.g., age, sex).

Exclusion Criteria:

* CT scans with poor image quality or severe artifacts that precluded accurate analysis.
* History of hip surgery or presence of internal fixation devices.
* Presence of bone tumors in the proximal femur.
* Severe hip deformity or prior fractures affecting the proximal femur.
* Pediatric patients or pregnant individuals (if applicable).

Where this trial is running

Beijing

• CT machine — Beijing, China (Recruiting)

Study contacts

• Study coordinator: hanwen Cheng, M.D

Conditions Bone Aging; Osteoporosis Diagnosis