Research Article
Diagnosis of Retinal Diseases Based on Bayesian Optimization Deep Learning Network Using Optical Coherence Tomography Images
Table 6
Performance of DenseNet201.
| Class labels | Precision (%) | Recall (%) | F1-score (%) | Feature extractor | Fine tuner | Feature extractor | Fine tuner | Feature extractor | Fine tuner |
| AMD | 97.34 | 100.00 | 94.82 | 100.00 | 95.12 | 99.12 | CNV | 91.08 | 99.82 | 92.45 | 98.12 | 92.91 | 89.66 | DME | 94.26 | 99.55 | 89.12 | 99.34 | 90.72 | 99.61 | CSR | 97.18 | 99.11 | 93.29 | 97.56 | 94.61 | 98.55 | DR | 89.99 | 99.79 | 96.51 | 98.69 | 94.95 | 97.99 | Drusen | 91.73 | 99.51 | 93.21 | 98.88 | 92.91 | 97.99 | MH | 91.00 | 100.00 | 96.57 | 99.11 | 93.98 | 98.44 | Normal | 92.09 | 99.91 | 94.12 | 98.99 | 94.01 | 98.03 | Macro average | 93.08 | 99.71 | 93.76 | 98.84 | 93.65 | 97.42 | Weighted average | 93.08 | 99.71 | 93.76 | 98.84 | 93.65 | 97.42 |
|
|