Table 5.
Literature review of SCZ diagnosis using ML and DL methods.
| Reference | Model | Year | Modality | Subjects | Training Set/Test Set | Accuracy (%) |
|---|---|---|---|---|---|---|
| [117] | SVM, RF, NB | 2020 | T1-w | 48 SCZ, 24 HC |
10-fold Cross-Validation | 68.6 |
| [118] | SVM | 2018 | fMRI | 21 SCZ, 54 HC |
10-fold Cross-Validation | 92.1 |
| [119] | SVM | 2017 | rs-fMRI | 86 SCZ, 84 HC |
10-fold Cross-Validation | 72 |
| [120] | ANN | 2015 | T1-w | 198 SCZ, 191 HC |
10-fold Cross-Validation | - |
| [121] | ANN | 2015 | rs-fMRI | 50 SCZ, 50 HC |
5-fold Cross-Validation | 85.8 |
| [122] | 2D-CNN | 2021 | T1-w | 500 HC slices, 500 SCZ slices |
- | 94.33 |
| [123] | SAE | 2019 | T1-w | 35 SCZ, 40 HC |
10-fold Cross-Validation | - |
| [124] | RNN | 2019 | fMRI | 558 SCZ, 542 HC |
5-fold Cross-Validation | 83 |
| [125] | Transfer learning | 2019 | rs-fMRI | 151 SCZ, 160 HC |
8/1 | - |
| [126] | SAE | 2016 | fMRI | 72 SCZ, 74 HC |
10-fold Cross-Validation | 92 |
| [127] | SAE | 2018 | rs-fMRI | 357 SCZ, 377 HC |
5-fold Cross-Validation | 85 |
| [128] | 3D-CNN | 2019 | rs-fMRI | 72 SCZ, 74 HC |
10-fold Cross-Validation | 98.09 |
| [129] | AE | 2016 | T1-w, fMRI |
69 SCZ, 75 HC |
- | - |