Table 6 Proposed algorithm in comparison with the previous.
Previous works | Maps used | Features extraction method | Classifier | Accuracy |
|---|---|---|---|---|
Dyrba et al.27 | FA and MD indices | Using the plant’s approach | SVM | 80% FA 83% MD |
Ahmed et al.29 | The visual appearance of MD maps | CHF | SVM | 86.7% AD vs. NC 73% AD vs. MCI |
Ahmed et al.31 | MD maps | A multimodal approach using LG-CHF | SVM | 90.2% AD vs. NC 77% AD vs. MCI |
The proposed CAD method | The visual appearance of MD, FA, and RD maps Also Fusion of features of the three maps | BoW based on SIFT and SURF features | SVM | 87.5% MD & FA multiclass 92.5% & 89% FA and MD AD vs. MCI 99 & 91% MD and FA AD vs. NC 97.5% Fusion features AD vs. NC, 89% RD multiclass, 98.5% RD AD vs. MCI |
