Automated Deep Learning Platform for Early Osteoporosis Risk Assessment from Dental CBCT Scans (OsteoXplore)

Osteoporosis is a skeletal disorder characterized by reduced bone mineral density resulting in impaired bone strength and increased risk of fractures. According to the WHO, it is the second most prevalent non-communicable disease after cardiovascular disorders. Osteoporotic fractures lead to disability and significantly reduce quality of life making early diagnosis, prevention, and timely treatment a major public health priority. Dual-energy X-ray absorptiometry, the current diagnostic gold standard, has limited accessibility and is not suitable for population-wide screening. Since postmenopausal women frequently undergo dental examinations, it has been hypothesized that dental imaging could support osteoporosis risk assessment. Through collaboration between dental and computer science experts, we have identified an optimal mandibular region in cone-beam computed tomography (CBCT) scans where cortical bone thickness serves as a reliable biomarker for osteoporosis risk in women. Using deep learning methods, we have developed three sequential automated modules for CBCT image analysis: (1) classification of relevant slices, (2) localization of anatomical reference points, and (3) cortical bone thickness quantification. This interdisciplinary project aims to integrate these modules into a unified AI-driven platform – OsteoXplore – perform external validation, develop a business model and analyze intellectual property rights and potential patentability.