Table 1. Quantitative evaluation for pediatric cataract classification and grading results with different CAD models.
| Metrics | WT | LBP | SIFT | COTE | CNN | |
|---|---|---|---|---|---|---|
| Classification | ACC (%) | 88.26 (0.02) § | 85.10 (0.02) | 77.76 (0.01) | 71.12 (0.13) | 97.07 (0.01) |
| SPC (%) | 95.38 (0.03) | 87.82 (0.01) | 88.03 (0.04) | 74.58 (0.39) | 97.28 (0.01) | |
| SEN (%) | 80.00 (0.06) | 81.95 (0.05) | 65.85 (0.03) | 67.06 (0.19) | 96.83 (0.02) | |
| Area grading | ACC (%) | 78.29 (0.03) | 70.72 (0.06) | 76.09 (0.04) | 74.17 (0.08) | 89.02 (0.01) |
| SPC (%) | 76.74 (0.12) | 56.40 (0.11) | 81.98 (0.04) | 59.88 (0.36) | 86.63 (0.06) | |
| SEN (%) | 79.41(0.06) | 81.07 (0.04) | 71.81(0.09) | 84.34(0.15) | 90.75(0.04) | |
| Density grading | ACC (%) | 83.39 (0.03) | 71.00 (0.04) | 77.32 (0.02) | 82.96 (0.05) | 92.68 (0.01) |
| SPC (%) | 82.69 (0.03) | 62.64 (0.09) | 83.83 (0.04) | 79.99 (0.13) | 91.05 (0.02) | |
| SEN (%) | 83.95 (0.07) | 77.49 (0.01) | 72.29 (0.05) | 85.28 (0.07) | 93.94 (0.02) | |
| Location grading | ACC (%) | 78.77 (0.03) | 76.09 (0.04) | 80.73 (0.01) | 67.83 (0.07) | 89.28 (0.03) |
| SPC (%) | 71.32 (0.08) | 56.65 (0.06) | 74.77 (0.09) | 52.83 (0.37) | 82.70 (0.06) | |
| SEN (%) | 83.08 (0.04) | 87.31 (0.04) | 84.23 (0.05) | 76.54 (0.23) | 93.08 (0.04) |
Footnotes: ACC (accuracy); SPC (specificity); SEN (sensitivity); WT (wavelet transformation); LBP (local binary pattern); SIFT (scale-invariant feature transform); COTE (color and texture features); CNN (convolutional neural network);
§Mean (Standard Deviation).