TABLE 4.

Testing accuracy, precision, recall, and F1-score for all class label.

Model	Class label	Accuracy	Precision	Recall	F1-score
CNN from scratch	AD	0.7244	0.7775	0.7244	0.7500
	CN	0.7132	0.7150	0.7132	0.7141
	MCI	0.7433	0.6941	0.7433	0.7178
VGG-16	AD	0.8390	0.8249	0.8390	0.8319
	CN	0.8137	0.7394	0.8137	0.7748
	MCI	0.7042	0.7978	0.7042	0.7481
ResNet-50	AD	0.8317	0.7715	0.8317	0.8005
	CN	0.7500	0.7286	0.7500	0.7391
	MCI	0.6895	0.7726	0.6895	0.7287

Bold values are highest value.