Abstract
Background: As the gold standard for clinical osteoporosis diagnosis, bone mineral density has significant limitations in bone strength assessment and fracture risk prediction. The purpose of this study is to explore a new osteoporotic bone quality evaluation criteria from both diagnosis site selection and bone strength prediction. Methods: Ovariectomized rats with different intensity swimming therapy were investigated in this study. The lumbar vertebrae and femurs of all the rats were scanned by synchrotron radiation computed tomography. Bone microstructure analysis and finite element analysis were combined to obtain bone microstructure parameters and estimated bone strength. And the sensitivity of different skeletal sites to therapy was explored. An elastic network regression model was established to predict bone strength by integrating additional bone microstructure characteristics besides bone mass.Results: Histomorphometry analysis showed that swimming therapy could reduce the risk of osteoporosis of lumbar vertebrae and femur and suggested that the femur might be a more suitable site for osteoporosis diagnosis and efficacy evaluation than the lumbar vertebrae. The average coefficient of determination and average root mean squared error of our predictive model were 0.774 and 0.110. Bland-Altman analysis showed that our model could be a good alternative to the finite element method. Conclusions: The present study developed a machine learning model for prediction of bone strength of osteoporosis model based on synchrotron x-ray imaging and demonstrated that different skeletal sites had different sensitivity to therapy, which is of great significance for the early diagnosis of osteoporosis, the prevention of fractures and the monitoring of therapy.
Screening and validation of serum protein biomarkers for early postmenopausal osteoporosis diagnosis
