Table 7 Performance comparison of proposed hybrid models with existing literature on ADNI dataset
Authors | Methods | Dataset | No of Images | Accuracy in % |
|---|---|---|---|---|
Weiming Lin et al. [29] | CNN | AD vs NC | 818 | 88.79 |
Kanghan et al [18] | Convolutional Autoencoder +Transfer learning | AD vs NC | 694 | 86.60 |
P C Muhammed Raees et al [22] | DNN | NC vs.MCI vs.AD | - | 80–90 |
Proposed HLM1 | 3D CNN→ML Model [stage 1] ML Model [stage 2] | NC vs Symptomatic AD MCI vs AD | 630 | 94.4 93.6 |
Proposed HLM2 | 3D CNN→Fuzzy Clustering [stage 1] Extra Tree Classifier [stage 2] | NC vs Symptomatic AD MCI vs AD | 630 | 92.8 93.8 |