Table 3 Comparison of RMSE and MAE for MMSE, CDR and Age prediction tasks.
Task | Reference | Methods | RMSE | MAE | MAPE | CC |
|---|---|---|---|---|---|---|
MMSE | Baron et al. (Voxel)46 | GM Density + SVR | 2.730 | – | – | 0.309 |
Zhang et al. (ROI)47 | ROI + SVR | 2.782 | – | – | 0.306 | |
Zhang et al.(LMF)48 | Landmark + SVR | 2.754 | – | – | 0.331 | |
Liu et al. (DM2L)49 | Multi-Channel | 2.373 | – | – | 0.567 | |
Liu et al.(wiseDNN)21 | Weakly Supervised | 2.401 | – | – | 0.525 | |
Duc et al.( LLSR)20 | gICA + SVM-RFE | 3.270 | – | – | – | |
Proposed method | 3D CNN | 2.303 | 1.973 | 7.77% | 0.601 | |
CDR | Yang et al.19 | SVR | – | 0.232 | – | 0.715 |
CL | – | 0.208 | – | 0.716 | ||
DPN | – | 0.169 | – | 0.805 | ||
Proposed method | 3D CNN | 0.213 | 0.159 | Â | 0.799 | |
Age | Cole et al.50 | 3D CNN | 5.31 | 4.16 | – | 0.96 |
Jonsson et al.51 | 3D ResNet | – | 4.641 | – | – | |
Mouches et al.52 | 3D SFCN | – | 3.85 | – | – | |
Cai et al.38 | Graph Transformer | 5.04 | 3.91(CN) | – | 0.748 | |
9.50 | 7.77(AD) | – | 0.155 | |||
Proposed method | 3D CNN | 4.535 | 3.81 | 4.983% | 0.602 |
