Inference of bone density from X-ray using AI methods

Osteoporosis and the associated low-impact fracture is a global problem, particularly in the aging population. Patients with a femoral neck fractures had a higher mortality rate; for patients older than 65 years, 20% of them died within one year from fracture. The prevalence of osteoporosis among people older than 50 years is estimated at approximately 30%; however, only 10% of them are diagnosed. Furthermore, only 27% of the people who had fractures undergone bone density testing. These data suggest a large unmet medical need in the detection of osteoporosis. The gold standard modality to measure bone mineral density (BMD) is the dual-energy x-ray absorptiometry (DXA). However, cost and accessibility prevents its large scale population screening. In contrast, plain pelvis film is standard x-ray which is of low cost and has a much higher population coverage and accessibility. Therefore, we developed a deep learning algorithm that can detect bone density directly by conventional pelvic
+ Inquiry