Table 2 Multi-label tooth diagnosis classification metrics. The results on the held-out test split are averaged for five models trained with 5-fold cross-validation. See subsection 3.4 for elaboration on the metrics
From: Combining public datasets for automated tooth assessment in panoramic radiographs
Accuracy | F1-score | AUC | |
|---|---|---|---|
Early Caries | 0.843 | 0.662 | 0.881 |
Deep Caries | 0.969 | 0.683 | 0.960 |
Periapical Lesion | 0.987 | 0.603 | 0.974 |
Impacted | 0.988 | 0.901 | 0.996 |