| Popis: |
Purpose:This study aims to explore a radiomics model based on lumbar spine CT imaging for detecting osteoporosis. Method:A retrospective analysis of 170 patients who underwent both CT and dual x-ray absorptiometry (DXA) of the lumbar spine in our hospital.According to the DXA results of these patients, 50 patients were normal, 50 patients had osteopenia, and 70 patients had osteoporosis.A total of 851 radiomic features were extracted from the images of the segmentation images in the lumbar CT imaging and the most significant radiomic features were selected to establish fourclassification models: normal vs osteoporosis , normal vs osteopenia, osteopenia vs osteoporosis，normal vs abnormal patients.The estimated area under the receiver operating characteristic curve（ROC）was used to assess the effectiveness of the classification models. Results :The area under the curve (AUC) was 0.942 (95% CI, 0.919-0.964) and 0.933 (95% CI, 0.896-0.971) for the model of normal vs.osteoporosis in the training and validation cohorts, respectively; 0.738 (95% CI, 0. 683-0.792) and 0.749 (95% CI, 0.663- 0.835) for the model of normal vs. Osteopenia ;0.822 (95% CI, 0.778-0.865) and 0.777 (95% CI, 0.700-0.853) for the model of osteopenia vs. Osteoporosis；0.867(95% CI, 0.833-0.900) and 0.825 (95% CI, 0.768-0.882) for the model of normal vs. abnormal. Conclusion: Radiomics based on lumbar CT imaging has great application value in diagnosing abnormal bone mineral density（BMD）. |
