Table 2 Results for each model doing deep tuning and 10-fold cross validation
From: CNNs for automatic glaucoma assessment using fundus images: an extensive validation
Model name | AUC | 95% confidence interval | Accuracy | Specificity | Sensitivity | Fscore | P-value |
|---|---|---|---|---|---|---|---|
VGG16 [15] | 0.9632 (0.0149) | 95.81–96.87% | 0.8948 (0.0253) | 0.8816 (0.0612) | 0.9057 (0.0331) | 0.9005 (0.0231) | 0.3240 |
VGG19 [15] | 0.9686 (0.0158) | 96.45–97.39% | 0.9069 (0.0318) | 0.8846 (0.0362) | 0.9240 (0.0434) | 0.9125 (0.0312) | 0.3607 |
InceptionV3 [30] | 0.9653 (0.0135) | 96.12–97.32% | 0.9000 (0.0201) | 0.8752 (0.0358) | 0.9216 (0.0311) | 0.9056 (0.0236) | 0.3126 |
ResNet50 [32] | 0.9614 (0.0171) | 95.62–96.77% | 0.9029 (0.0249) | 0.8943 (0.0350) | 0.9105 (0.0282) | 0.9076 (0.0251) | 0.3885 |
Xception [33] | 0.9605 (0.0170) | 95.92–97.07% | 0.8977 (0.0264) | 0.8580 (0.0398) | 0.9346 (0.0247) | 0.9051 (0.0274) | 0.2729 |