A tool for osteoporosis opportunistic screening v1.0
Research type
Research Study
Full title
A tool for osteoporosis opportunistic screening
IRAS ID
359237
Contact name
Yilin Zhang
Contact email
Sponsor organisation
University of Southampton
Duration of Study in the UK
2 years, 6 months, 12 days
Research summary
This research aims to explore the use of X-ray images to measure Bone Mineral Density (BMD), a key indicator of bone health. BMD helps in diagnosing conditions like osteoporosis, a disease that weakens bones and increases the risk of fractures. Currently, BMD is commonly measured using expensive and time-consuming methods such as dual-energy X-ray absorptiometry (DXA).
The study will develop an AI model that analyzes hip X-ray images to predict a person’s BMD. AI systems will be trained on a dataset of X-ray images and corresponding BMD values, allowing the model to learn how to estimate BMD from these images. By using AI to assess BMD, this research aims to improve the speed and accuracy of bone health assessments, making it easier for doctors to detect conditions like osteoporosis and other bone-related diseases at an earlier stage.
This research is important because it could make BMD assessments more accessible and efficient. With AI, doctors could quickly and accurately determine a patient's bone density from X-ray images, enabling earlier diagnosis of conditions such as osteoporosis. Early detection can lead to better treatment options, reduce the risk of fractures, and ultimately improve patient outcomes.
This project will use data from the Hertfordshire Cohort Study. Ethical approval was granted under ERGO 12827 (REC Ref: 11/EE/0196 / IRAS 80524) – Hertfordshire Bone Studies (HBS 2011–12), which permits the dataset to be stored for up to 20 years.The dataset also overlaps with data from the EPOSA study (European Project on Osteoarthritis), which was previously approved under ERGO 18456 (end date: 31 December 2019), confirming that participants consented to health and ageing research.
The development of the AI model will take several months. The participants' data will be used for training the model, and there will be no direct involvement or additional procedures for participants during this period.
We as an organisation - acting as sponsor - cannot confidently assure that re-identification of individuals is not possible. The reason for this, is that our organisation holds additional information that would allow linkage back to the same individuals.
REC name
HSC REC B
REC reference
25/NI/0124
Date of REC Opinion
18 Aug 2025
REC opinion
Favourable Opinion