Evaluation of deep learning-based quantitative computed tomography for opportunistic osteoporosis screening
Sangseok Oh1, Woo Young Kang1, Heejun Park1
1Department of Radiology, Guro Hospital, Korea University Medical Center, Korea University College of Medicine, 148, Gurodong-ro, Guro-gu, Seoul, 08308, Republic of Korea.
Scientific Reports
|January 5, 2024
Summary
Deep learning automated bone mineral density (BMD) measurement using routine CT scans shows high accuracy for osteoporosis screening. This method offers a reliable alternative for opportunistic BMD assessment, improving diagnostic efficiency.
Area of Science:
- Radiology
- Artificial Intelligence
- Osteoporosis Research
Background:
- Osteoporosis diagnosis typically relies on dual-energy X-ray absorptiometry (DXA), which is not always opportunistic.
- Routine computed tomography (CT) scans offer a potential avenue for opportunistic bone mineral density (BMD) assessment.
Purpose of the Study:
- To evaluate the diagnostic efficacy of a deep learning (DL)-based automated BMD measurement for opportunistic osteoporosis screening using routine CT scans.
Main Methods:
- A DL-based automated quantitative CT (DL-QCT) solution was assessed on 112 routine CT scans.
- DL-BMD measurements from L1 and L2 vertebrae were validated against manual BMD (m-BMD) and central DXA.
- Receiver operating characteristic (ROC) curve analysis determined diagnostic performance for low BMD and osteoporosis.
Main Results:
- DL-BMD demonstrated excellent concordance with m-BMD (r=0.961/0.979).
- ROC analysis showed comparable diagnostic performance to central DXA for low BMD (AUC=0.847) and osteoporosis (AUC=0.770).
- DL-BMD achieved high sensitivity, specificity, and accuracy when compared to m-BMD for both low BMD and osteoporosis detection.
Conclusions:
- DL-based automated BMD measurement on routine CT scans exhibits excellent agreement with manual measurements.
- This DL-QCT approach shows significant potential for accurate and opportunistic osteoporosis screening.
- The findings suggest DL-BMD can effectively detect low BMD and osteoporosis, complementing conventional methods.
