Table 3.

Comparison with the state-of-the-art methods for MCI conversion prediction

Method	Modalities	#Subjects	Performances
			ACC (%)	SEN (%)	SPE (%)	AUC
Duchesne and Mouiha 2011	MRI	20 MCI-C, 29 MCI-NC	72.3 %	75 %	62 %	0.794
Hinrichs et al. 2011	MRI, FDG-PET, CSF, APOE	119 MCI	N/A	N/A	N/A	0.7911
Davatzikos et al. 2011	MRI, CSF	69 MCI-C, 170 MCI-NC	61.7 %	95 %	38 %	0.734
Zhang et al. 2012a	MRI, FDG-PET, CSF	38 MCI-C, 50 MCI-NC	78.4 %	79 %	78 %	0.768
Coupé et al. 2012c	MRI	167 MCI-C, 238 MCI-NC	71 %	70 %	72 %	N/A
Wee et al. 2013	MRI	89 MCI-C, 111 MCI-NC	75.05 %	N/A	N/A	0.8426
Westman et al. 2012	MRI, CSF	81 MCI-C, 81 MCI-NC	68.5 %	74.1 %	63 %	0.76
Zhang et al. 2012b	MRI, FDG-PET, CSF	43 MCI-C, 48 MCI-NC	73.9 %	68.6 %	73.6 %	0.797
Cho et al. 2012	MRI	72 MCI-C, 131 MCI-NC	71 %	63 %	76 %	N/A
Eskildsen et al. 2013	MRI	161 MCI-C, 227 MCI-NC	75.4 %	70.5 %	77.6 %	0.82
Young et al. 2013	MRI, FDG-PET, CSF, APOE	47 MCI-C, 96 MCI-NC	74.1 %	78.7 %	65.6 %	0.795
Proposed method	MRI, FDG-PET, CSF	43 MCI-C, 56 MCI-NC	80.1 %	85.3 %	73.3 %	0.852