Table 5.
Classification performance for the proposed method compared with published state-of-the art methods for differentiating between MCIs vs. MCIc.
| Method | Modality | Subjects | AUC | ACC | SEN | SPEC |
|---|---|---|---|---|---|---|
| Zhang et al. (2012) | Longitudinal (MRI + PET) | 88 | 76.8 | 78.4 | 79 | 68 |
| Young et al. (2013) | MRI + PET + APOE | 143 | 79.5 | 69.9 | 78.7 | 65.6 |
| Suk et al. (2014) | PET + MRI | 204 | 74.66 | 75.92 | 48.04 | 95.23 |
| Cheng et al. (2015) | MRI + PET + CSF | 99 | 84.8 | 79.4 | 84.5 | 72.7 |
| Moradi et al. (2015) | MRI + AGE + Cognitive measure |
264 | 90.20 | 81.72 | 86.65 | 73.64 |
| Beheshti et al. (2017) | MRI | 136 | 75.08 | 75 | 76.92 | 73.23 |
| Liu et al. (2017) | MRI + PET | 234 | 80.8 | 73.5 | 76.19 | 70.37 |
| Long et al. (2017) | MRI (AMYG) | 227 | 93.2 | 88.99 | 86.32 | 90.91 |
| Proposed method | MRI + PET + CSF + APOE genotype |
82 | 93.59 | 94.86 | 100 | 88.71 |