Table 4.
AUC, accuracy, sensitivity, specificity and F1 score of four classification cases obtained from the radiomics-based and trabecular parameters-based model
| Normal vs. OA | Mild vs. advanced OA | Normal vs. advanced OA | Normal vs. mild OA | |||||
|---|---|---|---|---|---|---|---|---|
| Radiomics | Trabecular | Radiomics | Trabecular | Radiomics | Trabecular | Radiomics | Trabecular | |
| AUC |
0.961 (0.912–1.000) |
0.873 (0.788–0.957) |
0.995 (0.975–1.000) |
0.691 (0.551–0.830) |
0.997 (0.982–1.000) |
0.918 (0.846–0.990) |
0.919 (0.847–0.995) |
0.751 (0.628–0.874) |
| Accuracy | 0.920 | 0.765 | 0.964 | 0.625 | 0.984 | 0.820 | 0.898 | 0.727 |
| Sensitivity | 0.982 | 0.911 | 0.966 | 0.690 | 1.000 | 0.828 | 0.889 | 0.518 |
| Specificity | 0.812 | 0.562 | 0.963 | 0.556 | 0.969 | 0.812 | 0.906 | 0.750 |
| F1 score | 0.919 | 0.774 | 0.964 | 0.623 | 0.984 | 0.820 | 0.898 | 0.639 |
AUC Area under the curve. AUCs were given with the 95% confidence interval