Table 4.
Diagnostic performance of three predictive models for predicting invasive adenocarcinoma lesions.
| AUC 1 | AUC 2 | AIC | BIC | Sensitivity | Specificity | |
|---|---|---|---|---|---|---|
| Model 1 Clinical model (diameter, spiculation sign, pleural traction sign) |
0.833 | 0.852 | 699.102 | 717.998 | 0.759 | 0.791 |
| Model 2 Histogram model (variance, entropy, the mean density of the solid component, volume of the solid component) |
0.833 | 0.892 | 694.037 | 717.656 | 0.822 | 0.849 |
| Model 3 Integrated model (Clinical model+ Histogram model) |
0.833 | 0.886 | 673.056 | 710.847 | 0.830 | 0.814 |
AUC 1, 10-fold AUC result based on training cohort; AUC 2, results based on validation cohort. AIC, Akaike Information Criterion; BIC, Bayesian Information Criterion.