Table 2.

Classification results obtained on the test set in other works in literature. All the selected works had employed the ADNI dataset, except Etminani et al. (2021) [15] which integrated the ADNI dataset with data of the European-Dementia of Lewy Body (DLB) Consortium (E-DLB)

Author	Classification Task	Data Collection	Test Set	Accuracy	Sensitivity	Specifity	AUC	Precision	F1 score	MCC
Cheng and Lui [9]	AD vs. CN	339 subjects from ADNI	10% (10-fold validation)	0.91	0.91	0.91	0.95
	MCI vs. CN			0.79	0.78	0.80	0.84
Lu et al. [10]	AD vs. CN	1051 subjects from ADNI	10% (10-fold validation)	0.94	0.92	0.92
Zheng et al. [11]	AD vs. CN	962 images from ADNI	20% (5-fold validation)	0.91	0.86	0.95
Yee et al. [12]	AD vs. CN	596 subjects from ADNI	20% (5-fold validation)	0.93	0.92	0.94	0.98
Ding et al. [13]	AD vs. all	2109 images from 1002 subjects	10% (10-fold validation)		0.81	0.94		0.76	0.78
	MCI vs. all				0.54	0.68		0.55	0.55
	CN vs. all				0.59	0.75		0.60	0.59
Tufail et al. [14]	AD vs. all	90 images (Training + Validation)	23 images (12 CN, 7 MCI, 4 AD)	0.80	0.71	0.84	0.78	0.68	0.70	0.55
	MCI vs. all			0.60	0.35	0.72	0.53	0.38	0.36	0.68
	CN vs. all			0.74	0.65	0.79	0.72	0.62	0.63	0.43
Etminani et al. [15]	AD vs all	556 subjects from ADNI, 201 subjects from E-DLB	10% (73 cases)		0.91	0.92		0.83	0.87
	MCI vs. all				0.17	0.94		0.20	0.18
	DLB vs. all				0.86	1.00		1.00	0.92
	CN vs. all				0.88	0.90		0.81	0.84
Yiăit et al. [16]	AD vs CN	985 images from ADNI	20% (six-fold validation)	0.72
	MCI vs. CN			0.92