Table 4.
Comparison of classification accuracy with state‐of‐the‐art methods
| Methods | Dataset (AD/MCI/HC) | AD vs. HC (%) | MCI vs. HC (%) |
|---|---|---|---|
| Kohannim et al. | MRI + PET + CSF (40/83/43) | 90.70 | 75.80 |
| Walhovd et al. | MRI + CSF (38/73/42) | 88.80 | 79.10 |
| Hinriches et al. | MRI + PET + CSF + APOE + Cognitive scores (48/119/66) | 92.40 | n/a |
| Westman et al. | MRI + CSF (96/162/111) | 91.80 | 77.60 |
| Zhang and Shen | MRI + PET + CSF (45/91/50) | 93.30 | 83.20 |
| Gray et al. | MRI + PET (51/75/35) | 89.00 | 74.60 |
| Liu et al. | MRI+PET+CSF (51/99/52) | 94.37 | 78.80 |
| Suk et al. | MRI+PET+CSF+ Cognitive scores (51/99/52) | 95.90 | 85.00 |
| Proposed method | MRI+PET+CSF (51/99/52) | 97.12 | 87.09 |
Note. Numbers in parentheses denote the number of AD/MCI/NC subjects in the dataset used. Boldface denotes the best performance in each classification task.