Table 3 Main studies using deep convolutional network
From: Machine learning applied to retinal image processing for glaucoma detection: review and perspective
Methods | Year | Architecture | Metrics (%) | ||
|---|---|---|---|---|---|
Acc | Sp | Sn | |||
Li et al. [55] | 2018 | Inception-v3 | 92 | 95.6 | 92.34 |
Fu et al. [58] | 2018 | Disc-aware ensemble network (DENet) | 91.83 | 83.80 | 83.80 |
Raghavendra et al. [62] | 2018 | Eighteen-layer CNN | 98.13 | 98.3 | 98 |
dos Santos Ferreira et al. [63] | 2018 | U-net for segmentation and fully connected with dropout for classification | 100 | 100 | 100 |
Christopher et al. [65] | 2018 | ResNet50 | 97 | 93 | 92 |
Chai et al. [68] | 2018 | MB-NN | 91.51 | 92.33 | 90.90 |
Bajwa et al. [69] | 2019 | Four convolutional layers and fully connected layers | 87.40 | 85 | 71.17 |
Liu et al. [72] | 2019 | ResNet | 99.6 | 97.7 | 96.2 |