Table 4.
The baseline Dice score and Mean Absolute Error (with Standard Deviation) results of 4 class pixel segmentation with various neural network models. The best results are highlighted in bold.
| Model | Average Dice Score (SD) | MAE (SD) [px] | |||||
|---|---|---|---|---|---|---|---|
| Vitreous | Preretinal Space |
Retina | Region b. Retina |
PCV | ILM | RPE | |
| Graph search | 0.9842 (0.033) | 0.8216 (0.207) | 0.9385 (0.163) | 0.9888 (0.028) | 7.29 (13.7) | 5.10 (12.1) | 5.20 (15.5) |
| UNet | 0.9942 (0.015) | 0.9458 (0.088) | 0.9929 (0.002) | 0.9976 (0.003) | 2.83 (6.05) | 0.56 (0.14) | 0.75 (0.35) |
| LFUNet | 0.9928 (0.017) | 0.9337 (0.099) | 0.9922 (0.003) | 0.9972 (0.003) | 3.11 (5.72) | 0.60 (0.18) | 0.84 (0.46) |
| Attention UNet | 0.9822 (0.037) | 0.8679 (0.197) | 0.9918 (0.012) | 0.9953 (0.010) | 5.40 (10.04) | 0.63 (0.77) | 0.87 (1.11) |
| ReLayNet | 0.9529 (0.056) | 0.7827 (0.245) | 0.9906 (0.008) | 0.9814 (0.032) | 25.66 (26.34) | 0.84 (1.30) | 1.45 (1.77) |