Strut analysis for osteoporosis detection model using dental panoramic radiography
Jae Joon Hwang1, Jeong-Hee Lee1, Sang-Sun Han1
11 Department of Oral and Maxillofacial Radiology, Yonsei University College of Dentistry, Seoul, Republic of Korea.
Dento Maxillo Facial Radiology
|July 15, 2017
Summary
Osteoporosis can be detected using strut analysis of panoramic radiographs. This method, focusing on the endosteal margin, achieved high accuracy in identifying osteoporosis.
Area of Science:
- Dental radiology
- Medical imaging analysis
- Osteoporosis diagnostics
Background:
- Osteoporosis detection often relies on bone mineral density measurements.
- Panoramic radiography is a common dental imaging technique.
- Novel methods for osteoporosis detection using dental images are needed.
Purpose of the Study:
- To identify variables for osteoporosis detection using strut analysis, fractal dimension (FD), and gray level co-occurrence matrix (GLCM) on panoramic radiographs.
- To develop an osteoporosis detection model utilizing these image analysis techniques.
- To evaluate the efficacy of different regions of interest for detecting osteoporosis.
Main Methods:
- 454 panoramic radiographs were analyzed, equally divided between osteoporotic and non-osteoporotic patients.
- Strut features, FD, and GLCM were analyzed in specific regions, including the endosteal margin.
- A decision tree and support vector machine were employed to build and validate the osteoporosis detection model.
Main Results:
- The endosteal margin area revealed significant differences in FD, GLCM, and strut variables between patient groups.
- Strut variables in the endosteal margin area demonstrated high diagnostic performance (e.g., 97.2% accuracy with SVM).
- Combining strut variables with FD or GLCM did not improve diagnostic accuracy.
Conclusions:
- Analysis of strut features in the endosteal margin of panoramic radiographs shows promise for osteoporosis detection.
- Panoramic radiography-based models can potentially aid in osteoporosis screening.
- Further development of imaging analysis techniques for osteoporosis diagnosis is warranted.
