Back to the list
Congress: ECR25
Poster Number: C-15858
Type: Poster: EPOS Radiologist (scientific)
Authorblock: S. Park, J. Park, M. I. Kim, M. Kim, J. Song, S. J. Bae, N. Kim; Seoul/KR
Disclosures:
Saerom Park: Employee: Promedius Inc.
Junhyeok Park: Employee: Promedius Inc.
Minje Ian Kim: Employee: Promedius Inc.
Minjee Kim: Employee: Promedius Inc.
Jeongmin Song: Employee: Promedius Inc.
Sung Jin Bae: Nothing to disclose
Namkug Kim: Shareholder: Promedius Inc.
Keywords: Artificial Intelligence, Computer applications, Musculoskeletal bone, Conventional radiography, Computer Applications-Detection, diagnosis, Screening, Osteoporosis
Purpose Osteoporosis is a significant public health concern, particularly among the elderly, as it leads to decreased bone density and a substantially increased risk of fractures [1]. However, identifying osteopenia—an intermediate stage before progression to osteoporosis—at an early stage is crucial as it predisposes individuals to fractures if left undiagnosed and untreated [2]. Early detection of osteopenia allows for timely intervention and effective fracture prevention. Despite its importance, the gold standard for diagnosing osteoporosis and osteopenia, dual-energy X-ray absorptiometry (DXA), has...
Read more Methods and materials We applied the Multi-Diversity Consistency Self-distillation (MDCS) [3] framework to develop a deep learning model for classifying BMD into three categories—normal, osteopenia, and osteoporosis—based on chest X-ray (CXR) images. These categories were defined by DXA T-scores as follows: normal (T-score ≥ −1.0), osteopenia (−2.5 < T-score < −1.0), and osteoporosis (T-score ≤ −2.5). A total of 69,201 CXRs paired with DXA T-scores —collected on the same date—were obtained from a single hospital and served as the training dataset. All images...
Read more Results In this study, we evaluated the performance of a bone mineral density (BMD) classification model using CXRs obtained from three distinct clinical settings—a tertiary university hospital (Hospital A), a secondary healthcare facility (Hospital B), and a facility serving veterans (Hospital C). The internal dataset from Hospital A comprised 15.98% male patients with a mean age of 58.7±6.76 years, while the external dataset from the same institution included 11.02% male patients with a mean age of 59.01±6.65 years. Hospital B’s dataset...
Read more Conclusion Our deep learning model, utilizing CXR images for the classification of BMD into normal, osteopenia, and osteoporosis categories, demonstrated robust performance across diverse clinical settings. The results highlight its potential as an opportunistic screening tool for bone health. By utilizing widely performed CXRs, this method could enhance accessibility and reduce the number of undiagnosed osteoporosis and osteopenia cases.Further refinements are needed to optimize the model across different healthcare environments. Variations in image quality, patient positioning, and population characteristics should be...
Read more References MAI, Ha T., et al. Two-thirds of all fractures are not attributable to osteoporosis and advancing age: implications for fracture prevention. The Journal of Clinical Endocrinology & Metabolism, 2019, 104.8: 3514-3520. Fan Y, Li Q, Liu Y, et al. Sex- and Age-Specific Prevalence of Osteopenia and Osteoporosis: Sampling Survey. JMIR Public Health Surveill. 2024;10:e48947. Published 2024 Apr 5. doi:10.2196/48947 Zhao, Qihao, et al. "MDCS: More diverse experts with consistency self-distillation for long-tailed recognition." Proceedings of the IEEE/CVF International Conference on Computer Vision....
Read more
GALLERY