Table 6.
The performance comparison of our model with other state-of-the-art studies report in the literature using the ADNI database for prediction of AD vs. CN and pMCI vs. sMCI.
| Method | Test subjects | AD vs. CN | pMCI vs. sMCI | ||||||
|---|---|---|---|---|---|---|---|---|---|
| ACC | SEN | SPE | AUC | ACC | SEN | SPE | AUC | ||
| VBF | 137AD+76sMCI+ 134pMCI+162CN |
0.760 | – | – | – | 0.660 | |||
| SVM-Landmark | 154 AD+346 MCI +207 CN |
0.822 | 0.774 | 0.861 | 0.881 | – | – | – | – |
| LBM |
385AD+465sMCI+ 205pMCI+429CN |
0.822 | 0.774 | 0.861 | 0.881 | 0.686 | 0.395 | 0.732 | 0.636 |
| MLP-RNN | 198AD+229CN | 0.897 | 0.868 | 0.925 | 0.921 | ||||
| Whole-3DCNN | 50AD+77sMCI+ 43pMCI+61CN |
0.800 | – | – | 0.870 | 0.520 | – | – | 0.520 |
| Multi-3DCNN | 48AD+58CN | 0.850 | 0.880 | 0.900 | – | – | – | – | – |
| 3D-DenseNet | 97AD+233MCI +119CN |
0.889 | 0.866 | 0.808 | 0.925 | ||||
| wH-FCN | 385AD+465sMCI +205pMCI+429CN |
0.903 | 0.824 | 0.965 | 0.951 | 0.809 | 0.526 | 0.854 | 0.781 |
| Our model | 326AD++470sMCI +242pMCI+413CN |
0.911 | 0.888 | 0.914 | 0.950 | 0.801 | 0.520 | 0.856 | 0.789 |
The best results are highlighted in bold.