Prediction of Bone Mineral Density based on Computer Tomography Images Using Deep Learning Model
Jujia Li1, Ping Zhang2, Jingxu Xu3
1Medical Imaging Department, Hebei Medical University Third Hospital, Shijiazhuang, China, 1203460713@qq.com.
Gerontology
|November 11, 2024
Summary
A new deep learning model accurately measures bone mineral density (BMD) from CT scans, improving osteoporosis diagnosis. This AI tool aids in identifying osteoporosis, a growing concern in aging populations.
Area of Science:
- Radiology
- Artificial Intelligence
- Gerontology
Background:
- Population aging is a global challenge, increasing the prevalence of osteoporosis.
- Osteoporosis significantly impacts the health of older adults.
- Current osteoporosis diagnosis and public awareness are insufficient.
Purpose of the Study:
- To develop a deep learning model for automatic bone mineral density (BMD) measurement.
- To enhance the diagnostic rate of osteoporosis using artificial intelligence.
Main Methods:
- Utilized CT and quantitative CT (QCT) scans from 801 subjects (2,080 vertebral bodies).
- Developed a multistage deep learning model for vertebral body segmentation and BMD prediction.
- Validated model performance against QCT measurements using accuracy, precision, recall, and F1-score.
Main Results:
- The deep learning model demonstrated a strong correlation with QCT-measured BMD (R² = 0.95-0.97).
- Diagnostic accuracy of the model reached 0.88-0.91 across different datasets.
- BMD decreased with age in both males and females within the study cohort.
Conclusions:
- The developed deep learning model effectively measures vertebral BMD automatically.
- The model shows strong performance in predicting osteoporosis, addressing a critical diagnostic gap.
