Validation of a Deep Learning Tool for Opportunistic Osteoporosis Screening Using Routine Non-Contrast CT Scans

Prospective Clinical Validation Study of a Deep Learning Model for Opportunistic Osteoporosis Screening Based on Non-Contrast CT Scans

The goal of this clinical trial is to test if an artificial intelligence (AI) tool called DeepBMD can accurately identify people at high risk for osteoporosis using routine chest or abdomen CT scans. The main questions it aims to answer are:

1. Can the DeepBMD tool correctly identify people who have osteoporosis compared to the standard bone density test, dual-energy X-ray absorptiometry (DXA)?
2. Is it practical to use this AI tool in real-world hospital settings to find and contact high-risk patients? Researchers will use the DeepBMD tool to analyze existing CT scans. If the tool flags a patient as high risk, researchers will call them to invite them for a standard bone density test (DXA).

Participants will:

1. Have their existing chest or abdomen CT scan analyzed by the DeepBMD AI tool;
2. Receive a phone call from the research team if identified as high risk;
3. Visit the clinic for a free standard bone density test (DXA) if they agree to participate.

Study Overview

• Status: Active, not recruiting
• Conditions: Osteoporosis
• Intervention / Treatment: Diagnostic test: DeepBMD model for osteoporosis risk screening

• Study Type: Observational
• Enrollment (Estimated): 100

Contacts and Locations

Study Locations

• China
  • Hubei
    • Wuhan, Hubei, China, 430022
      • Union Hospital, Tongji Medical College, Huazhong University of Science and Technology

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

• Sampling Method: Non-Probability Sample

Study Population

Patients who underwent non-contrast CT at our hospital (Union Hospital, Tongji Medical College, Huazhong University of Science and Technology) and were identified as high-risk for osteoporosis by the DeepBMD model.

Description

Inclusion Criteria:

• Underwent non-contrast CT at our institution, with qualified image quality and no severe artifacts;
• Identified as high-risk for osteoporosis by the DeepBMD model;
• Had valid contact information available in the PACS, possessed normal cognitive and communication abilities, and was able to cooperate with telephone follow-ups and on-site examinations;
• Voluntarily participated in the study, was able to sign a written informed consent form on-site, and agreed to undergo DXA examination.

Exclusion Criteria:

• Severe spinal deformity, postoperative spinal internal fixation, malignant bone metastasis, or severe osteolytic lesions that may interfere with measurements;
• A confirmed diagnosis of osteoporosis with ongoing standardized treatment;
• Inability to be contacted, explicit refusal of follow-up, or inability to visit the hospital for informed consent signing and DXA examination.

Study Plan

How is the study designed?

Number of groups / cohorts

1

Cohorts and Interventions

Group / Cohort	Intervention / Treatment
High-risk patients for osteoporosis identified by DeepBMD model; Patients who underwent routine chest or abdominal CT scans and were identified as high risk for osteoporosis by the DeepBMD AI model. These participants will be contacted via telephone, invited to the clinic, and undergo a free DXA scan to verify bone mineral density.	Diagnostic test: DeepBMD model for osteoporosis risk screening; The DeepBMD model is applied to routine chest or abdominal CT scans to identify patients at high risk for osteoporosis. This is a non-invasive image analysis used solely for screening and recruitment purposes, not as a therapeutic intervention.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Diagnostic performance of DeepBMD model for osteoporosis screening	The diagnostic performance of the DeepBMD model will be evaluated by comparing its predictions against the gold standard Dual-energy X-ray Absorptiometry (DXA). Specifically, we will calculate the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and the Area Under the Receiver Operating Characteristic Curve (AUC) for identifying patients with osteoporosis.	Concurrent with the DXA validation visit following the CT analysis (within 7 days).

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Feasibility of the DeepBMD screening and recall workflow	It will be assessed by calculating the proportion of patients identified as high-risk by DeepBMD who successfully complete the telephone follow-up and undergo the confirmatory DXA scan within the scheduled timeframe. We will also record the reasons for refusal or loss to follow-up to evaluate the acceptability of this AI-driven screening pathway.	At the end of recruitment

Collaborators and Investigators

• Sponsor: Union Hospital, Tongji Medical College, Huazhong University of Science and Technology
• Investigators: Principal Investigator: Fan Yang, PhD, MD, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology

Publications and helpful links

General Publications

• Jang S, Graffy PM, Ziemlewicz TJ, Lee SJ, Summers RM, Pickhardt PJ. Opportunistic Osteoporosis Screening at Routine Abdominal and Thoracic CT: Normative L1 Trabecular Attenuation Values in More than 20 000 Adults. Radiology. 2019 May;291(2):360-367. doi: 10.1148/radiol.2019181648. Epub 2019 Mar 26.
• Wang P, She W, Mao Z, Zhou X, Li Y, Niu J, Jiang M, Huang G. Use of routine computed tomography scans for detecting osteoporosis in thoracolumbar vertebral bodies. Skeletal Radiol. 2021 Feb;50(2):371-379. doi: 10.1007/s00256-020-03573-y. Epub 2020 Aug 7.
• Smith AD. Screening of Bone Density at CT: An Overlooked Opportunity. Radiology. 2019 May;291(2):368-369. doi: 10.1148/radiol.2019190434. Epub 2019 Mar 26. No abstract available.
• Zeng Q, Li N, Wang Q, Feng J, Sun D, Zhang Q, Huang J, Wen Q, Hu R, Wang L, Ma Y, Fu X, Dong S, Cheng X. The Prevalence of Osteoporosis in China, a Nationwide, Multicenter DXA Survey. J Bone Miner Res. 2019 Oct;34(10):1789-1797. doi: 10.1002/jbmr.3757. Epub 2019 Aug 29.
• Cheng X, Zhao K, Zha X, Du X, Li Y, Chen S, Wu Y, Li S, Lu Y, Zhang Y, Xiao X, Li Y, Ma X, Gong X, Chen W, Yang Y, Jiao J, Chen B, Lv Y, Gao J, Hong G, Pan Y, Yan Y, Qi H, Ran L, Zhai J, Wang L, Li K, Fu H, Wu J, Liu S, Blake GM, Pickhardt PJ, Ma Y, Fu X, Dong S, Zeng Q, Guo Z, Hind K, Engelke K, Tian W; China Health Big Data (China Biobank) project investigators. Opportunistic Screening Using Low-Dose CT and the Prevalence of Osteoporosis in China: A Nationwide, Multicenter Study. J Bone Miner Res. 2021 Mar;36(3):427-435. doi: 10.1002/jbmr.4187. Epub 2020 Nov 4.
• Lin X, Xiong D, Peng YQ, Sheng ZF, Wu XY, Wu XP, Wu F, Yuan LQ, Liao EY. Epidemiology and management of osteoporosis in the People's Republic of China: current perspectives. Clin Interv Aging. 2015 Jun 25;10:1017-33. doi: 10.2147/CIA.S54613. eCollection 2015.

Study record dates

Study Major Dates

• Study Start (Actual): July 28, 2026
• Primary Completion (Actual): August 28, 2026
• Study Completion (Estimated): September 1, 2026

Study Registration Dates

• First Submitted: July 22, 2026
• First Submitted That Met QC Criteria: July 27, 2026
• First Posted (Actual): July 31, 2026

Study Record Updates

• Last Update Posted (Actual): September 3, 2026
• Last Update Submitted That Met QC Criteria: September 1, 2026
• Last Verified: September 1, 2026

More Information

Terms related to this study

• Keywords: Osteoporosis; Deep Learning; Bone Mineral Density; Computed Tomography (CT); Opportunistic Screening
• Additional Relevant MeSH Terms: Bone Diseases; Musculoskeletal Diseases; Metabolic Diseases; Bone Diseases, Metabolic; Nutritional and Metabolic Diseases; Osteoporosis
• Other Study ID Numbers: UHCT260668

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: YES
• IPD Plan Description: De-identified individual participant data (IPD) will be made available to researchers who provide a methodologically sound proposal. The shared data will include the demographic information, DeepBMD screening results, and confirmatory DXA T-scores used in the study analyses. Requests should be directed to the corresponding author via email. Data will be available for non-commercial academic research purposes only. Applicants must sign a data access agreement prior to receiving the data.
• IPD Sharing Time Frame: Data will be available beginning 3 months following article publication and ending 36 months following article publication.
• IPD Sharing Access Criteria: Researchers who provide a methodologically sound proposal for specific research questions related to osteoporosis screening or AI diagnostics will be granted access. Approved researchers will have access to the de-identified dataset containing patient demographics, imaging analysis results, and clinical outcomes. Access will be granted via secure email transfer after signing a data use agreement.
• IPD Sharing Supporting Information Type: STUDY_PROTOCOL; SAP; ANALYTIC_CODE

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No