Research Article
A Novel Context Aware Joint Segmentation and Classification Framework for Glaucoma Detection
Table 6
Classification results under ablation study (No. of ResNeXt blocks is reduced to 3 from 4).
| | Dataset | Accuracy | Sensitivity | Specificity | Precision | FPR | F1 | MCC |
| | ACRIMA | 0.9453 | 0.9516 | 0.9418 | 0.9433 | 0.0582 | 0.9486 | 0.9456 | | rishti-gs | 0.9292 | 0.9344 | 0.9174 | 0.9344 | 0.0826 | 0.9374 | 0.9068 | | REFUGE | 0.9433 | 0.9480 | 0.9427 | 0.9029 | 0.0573 | 0.9278 | 0.9255 | | RIMOne-V1 | 0.9424 | 0.9476 | 0.9399 | 0.9298 | 0.0601 | 0.9418 | 0.9378 |
|
|