Table 2.
Diagnostic performance of the clinical factor model, radiomics signature, radiomics nomogram, and one experienced radiologist for detection of spinal metastasis
| Set | Model | AUC(95%CI) | Sensitivity(%) | Specificity(%) | Accuracy(%) |
|---|---|---|---|---|---|
| Training set | Clinical model | 0.715(0.641–0.791) | 87.0 | 54.3 | 76.7 |
| Radiomics signature | 0.847(0.795–0.898) | 75.0 | 73.9 | 74.7 | |
| Radiomics nomogram | 0.856(0.804–0.907) | 73.0 | 76.1 | 74.0 | |
| Radiologist | 0.711 (0.630–0.783) | 77.0 | 65.2 | 73.3 | |
| Validation set | Clinical model | 0.692(0.531–0.730) | 77.3 | 47.6 | 67.7 |
| Radiomics signature | 0.743(0.668–0.861) | 81.8 | 71.4 | 78.5 | |
| Radiomics nomogram | 0.853(0.764–0.919) | 84.1 | 85.7 | 84.6 | |
| Radiologist | 0.613 (0.484–0.731) | 75.0 | 47.6 | 66.2 | |
| External test set | Clinical model | 0.540(0.517–0.651) | 86.5 | 30.6 | 59.4 |
| Radiomics signature | 0.754(0.595–0.748) | 71.2 | 63.3 | 67.3 | |
| Radiomics nomogram | 0.762(0.605–0.751) | 78.8 | 67.1 | 68.3 | |
| Radiologist | 0.632 (0.531–0.726) | 67.3 | 59.2 | 63.4 |
AUC = area under the curve; CI = confidence interval