Table 4.
Statistical study of PD detection with handwriting images using ILN-GNet over traditional methods for dataset 1.
| Methods | Mean | Minimum | Median | SD | Maximum |
|---|---|---|---|---|---|
| XGBoost | 0.824 | 0.796 | 0.823 | 0.024 | 0.855 |
| LSTM-CNN | 0.912 | 0.883 | 0.911 | 0.024 | 0.944 |
| LinkNet | 0.919 | 0.891 | 0.916 | 0.024 | 0.952 |
| GhostNet | 0.829 | 0.800 | 0.827 | 0.025 | 0.863 |
| GoogleNet | 0.870 | 0.841 | 0.868 | 0.025 | 0.903 |
| AlexNet | 0.891 | 0.861 | 0.887 | 0.026 | 0.927 |
| EfficientNet | 0.770 | 0.742 | 0.766 | 0.026 | 0.806 |
| CNN | 0.839 | 0.810 | 0.837 | 0.024 | 0.871 |
| Ensemble | 0.810 | 0.780 | 0.806 | 0.026 | 0.847 |
| PROP | 0.950 | 0.921 | 0.947 | 0.025 | 0.984 |