Table 7.
Expert-outline-based two-stage classification performance.
| Database and number of images used | RDR-based assessment | CDR-based assessment | ||||
|---|---|---|---|---|---|---|
| Sensitivity | Specificity | Accuracy | Sensitivity | Specificity | Accuracy | |
| Drishti-GS (50) | 0.9706 | 0.9375 | 0.9600 | 0.9412 | 0.7500 | 0.8800 |
| Drions-DB (100) | 1.0000 | 0.9560 | 0.9600 | 1.0000 | 0.9121 | 0.9200 |
| Messidor (100) | 1.0000 | 0.9560 | 0.9600 | 0.8889 | 0.8791 | 0.8800 |
| Forus (126) | 0.9630 | 0.9697 | 0.9683 | 0.8889 | 0.9091 | 0.9048 |
| Bosch (60) | — | 1.0000 | 1.0000 | — | 1.0000 | 1.0000 |
| Consolidated (436 images) | 0.9625 | 0.9663 | 0.9656 | 0.9241 | 0.9104 | 0.9128 |