Table 8 Comparison of classification accuracy for recent studies.
classes | Authors | Target | Modality | Machine Learning | Brain Segmentation Method | Accuracy |
|---|---|---|---|---|---|---|
Two classes | Suk et al.18 | AD vs. HC | MRI + PET | Multi-Kernel SVM | 93 regions | 95.9% |
MCI vs. HC | 85.0% | |||||
MCI-C vs. MCI-NC | 75.8% | |||||
Ortiz et al.19 | AD vs. HC | FDG-PET + sMRI | SVM (Linear) | 42 subcortical regions | 92% | |
MCI vs. AD | 84% | |||||
HC vs. MCI | 86% | |||||
Li et al.20 | AD vs. HC | MRI + PET | RBM and SVM | 93 volumetric regions | 91.4% | |
MCI vs. HC | 77.4% | |||||
AD vs. MCI | 70.1% | |||||
MCI.C vs. MCI.NC | 57.4% | |||||
Khedher et al.21 | HC vs. AD | sMRI(T1) | SVM(Linear) | SPM8 | 87.12% | |
HC vs. MCI | 77.62% | |||||
MCI vs. AD | 85.41% | |||||
Our Method | AD vs. HC | fMRI | Linear-SVM | J-HCPMMP | 98.0% | |
MCI vs. AD | 92.0% | |||||
HC vs. MCI | 95.5% | |||||
Three classes | Quintana et al.22 | MCI vs. AD vs. HC | NPR | ANN | 55 regions | 66.67% |
Zhang et al.23 | MCI vs. AD vs. HC | MRI | SVM (RBF) | 66 volumetric features | 81.5% | |
Tong et al.24 | MCI vs. AD vs. HC | sMRI(T1) + PDG-PET + CSF + Genetics | NGF + SVM | 83 anatomical regions | 60.26% | |
Lama et al.25 | MCI vs. AD vs. HC | sMRI(T1) | PCA + RELM | FreeSurfer 5.3.0 | 61.58% | |
Son et al.26 | MCI vs. AD vs. HC | sMRI(T1) + rs-fMRI | Random Forest | 10 subcortical regions | 53.3% | |
Our Method | MCI vs. AD vs. HC | fMRI | Linear-SVM | J-HCPMMP | 88.0% |
