Table 5 Comparisons with other volume-based methods. The data with a number of subjects in each disease group consisted of: (CN, AD) for CN vs. AD and (CN, MCI) for CN vs. MCI task. The performance of the model is evaluated using the following metrics: balanced accuracy (BACC), sensitivity (SEN), specificity (SPE), and area under the curve (AUC).
From: Multimodal surface-based transformer model for early diagnosis of Alzheimer’s disease
Task | Methods | Modality | Data | Input size | BACC | SEN | SPE | AUC |
|---|---|---|---|---|---|---|---|---|
CN vs. AD | Li et al. (2018)8 | MRI | (229, 199) | (98 \(\times\) 78 \(\times\) 76) | 0.894 | 0.879 | 0.908 | 0.924 |
Cui and Liu (2019)9 | MRI | (223, 192) | (62 \(\times\) 48 \(\times\) 58) | 0.921 | 0.906 | 0.937 | 0.969 | |
Liu et al. (2020)11 | MRI | (119, 97) | (64 \(\times\) 48 \(\times\) 64) | 0.887 | 0.866 | 0.908 | 0.925 | |
Zhang et al. (2022)10 | FDG | (184, 146) | (128 × 128 × 128) | 0.972 | 0.960 | 0.985 | 0.967 | |
Qiu et al. (2024)36 | MRI, FDG | (317, 290) | (128 \(\times\) 128 \(\times\) 128) | 0.964 | 0.974 | 0.954 | 0.985 | |
Chen et al. (2024)37 | MRI, A\(\beta\) | (283, 144) | – | 0.936 | 0.890 | 0.982 | 0.970 | |
Zhang et al. (2023)25 | MRI | (360, 345) | – | 0.900 | 0.833 | 0.967 | 0.926 | |
Wee et al. (2019)16 | MRI | (654, 965) | – | 0.890 | 0.914 | 0.865 | – | |
Ours | MRI,A\(\beta\),Tau | (258, 55) | (40962 \(\times\) 5) | 0.937 | 0.889 | 0.984 | 0.943 | |
Ours | MRI, FDG | (101, 84) | (40962\(\times\) 4) | 0.962 | 0.976 | 0.948 | 0.969 | |
CN vs. MCI | Li et al. (2018)8 | MRI | (229, 403) | (98\(\times\) 78 \(\times\) 76) | 0.691 | 0.866 | 0.515 | 0.775 |
Cui and Liu (2019)9 | MRI | (223, 396) | (62 \(\times\) 48 \(\times\) 58) | 0.737 | 0.773 | 0.699 | 0.777 | |
Liu et al. (2020)11 | MRI | (119, 233) | (64\(\times\) 48 \(\times\) 64) | 0.746 | 0.795 | 0.698 | 0.775 | |
Zhang et al. (2022)10 | FDG | (184, 347) | (128 \(\times\) 128 \(\times\) 128) | 0.675 | 0.775 | 0.575 | 0.744 | |
Qiu et al. (2024)36 | MRI, FDG | (317, 506) | (128 \(\times\) 128 \(\times\) 128) | 0.736 | 0.730 | 0.730 | 0.761 | |
Chen et al. (2024)37 | MRI, A\(\beta\) | (283, 330) | – | 0.669 | 0.697 | 0.640 | 0.719 | |
Zhang et al. (2023)25 | MRI | (360, 613) | – | 0.698 | 0.891 | 0.504 | 0.738 | |
Wee et al. (2019)16 | MRI | (661, 1210) | – | 0.660 | 0.619 | 0.701 | – | |
Ours | MRI,A\(\beta\),Tau | (258, 159) | (40962 \(\times\) 5) | 0.759 | 0.684 | 0.833 | 0.764 | |
Ours | MRI, FDG | (101, 208) | (40962 \(\times\) 4) | 0.788 | 0.692 | 0.884 | 0.805 |
