Table 4.
Result comparison of our proposed model with up-to-date methods on the Messidor dataset for fundus classification. EEL stands for end-to-end learning.
| Model | Training Type | Method (%) | Process Type (%) | ACC (%) | AUC (%) | SEN (%) |
|---|---|---|---|---|---|---|
| Gulshan et al. [16] | CNN | Transfer Learning | Fundus Classification | - | 99.0 | 87.0 |
| Costa and Campilho [22] | SURF + CNN | EEL | Fundus Classification | - | 99.0 | - |
| Gargeya and Leng [17] | CNN | EEL | Fundus Classification | - | 94.0 | - |
| Wang et al. [23] | Zoom | EEL | Fundus Classification | 91.1 | 95.7 | - |
| Chen et al. [24] | SI2DRNet | EEL | Fundus Classification | 91.2 | 96.5 | 87.0 |
| WFDLN | CNN | Transfer Learning | Fundus Classification | 98.5 | 99.1 | 98.9 |