Research Article
Automatic Detection of AMD and DME Retinal Pathologies Using Deep Learning
Table 6
Comparison of the results with some state of the art methods.
| | CNN | Preprocessing | Accuracy |
| | Karri et al. | GoogLeNet | BM3D | 91.3% |
| | Wang et al. | VGG16 | FastNIMeans bilateralFilter | 91.6% | | InceptionV3 | FastNIMeans bilateralFilter | 92.7% | | VGG19 | FastNIMeans bilateralFilter | 98.2% |
| | Proposed architectures | Xception | ROI | 96.83% | | Xception+feature extraction from middle layer | ROI | 98.02% | | Inception-ResnetV2 | ROI | 93.43% | | Inception-ResnetV2+feature extraction from middle layer | ROI | 97.70% | | BCNN (Xception, Xception) | ROI | 97.84% | | BCNN (Inception-ResnetV2, Inception-ResnetV2) | ROI | 95.55% | | The from scratch “OCTorch-Net” | Without preprocessing | 99.68% |
|
|
