Table 3.
Summary of parameters yielding the most accurate logistic regression prediction from all lymph nodes that overlap in DWI, DCE-MRI, and PET image field-of-view.
| Parameters | Threshold | Accuracy | Sensitivity | Specificity | AUC |
|---|---|---|---|---|---|
| Ki | 0.18 | 0.96 | 1 | 0.95 | 0.98 |
| k3 | 0.43 | 0.94 | 0.82 | 0.97 | 0.97 |
| Volume | 0.32 | 0.90 | 0.64 | 0.97 | 0.89 |
| k3, k1 | 0.45 | 0.98 | 0.91 | 1 | 0.99 |
| K3, k2 | 0.36 | 0.98 | 0.91 | 1 | 0.98 |
| ADC, Ki | 0.18 | 0.96 | 1 | 0.95 | 0.98 |
| Ktrans, Ki | 0.18 | 0.96 | 1 | 0.95 | 0.98 |
| ve, Ki | 0.13 | 0.96 | 1 | 0.95 | 0.98 |
| ve, k3 | 0.15 | 0.96 | 1 | 0.95 | 0.99 |
| Fp, Ki | 0.18 | 0.96 | 1 | 0.95 | 0.98 |
| Ki, k3 | 0.68 | 0.96 | 0.82 | 1 | 0.98 |
| Ki, k2 | 0.18 | 0.96 | 1 | 0.95 | 0.99 |
| k3, Volume | 0.54 | 0.96 | 0.82 | 1 | 0.96 |
The table includes the threshold used to achieve the highest accuracy.