Table 1.

Summary of Related Works.

References	Deep Learning Model	Brain Region	Contribution	Limitations
[30]	VGG 19	Gray Matter	Model provided clear distinction between EMCI and LMC	Low classification accuracy for intermediate stages
[22]	CNN with Ensemble Learning	Whole brain	Baseline	Low classification accuracy for MCIc vs. MCI nc
[31]	3D Convolutional LSTM	Whole brain	Accuracy was improved	Prodromal stages of AD were not considered
[25]	3D CNN	Whole brain	Model learned temporal and spatial features	Model was designed for a specific size of fMRI volume
[28]	CNN with residual connection	Hippocampus	Baseline	N/A
[29]	Inception-ResNet V2	Whole brain	Based on various age groups, the model was able to extrapolate the prediction of different phases	Model did not address the problem of EMCI vs. NC binary classification
[33]	CNN	Whole brain	Model recognized pattern of brain functional changes associated with AD progression	Intermediate stages of AD were not considered
[36]	AlexNet	Whole brain	Model classified all stages of AD	Low binary classification accuracy
[37]	VGG16	Whole brain	Model was able to extract useful features the binary classification tasks	High computational complexity