Table 4 Performance metrics evaluated for four different image types used for the classification of ROP Stage with EfficientNetB4 on Validation set.
From: Automated diagnosis of plus form and early stages of ROP using deep learning models
Image type | Accuracy | F1 score | Sensitivity | Specificity |
|---|---|---|---|---|
Original image | 0.94 | 0.91 | 0.93 | 0.98 |
Mask | 0.86 | 0.82 | 0.86 | 0.95 |
Four-channel image | 0.91 | 0.91 | 0.91 | 0.96 |
Original image + Mask | 0.98 | 0.97 | 0.96 | 0.9925 |
