Significance of Normalization on Anatomical MRI Measures in Predicting Alzheimer’s Disease

Table 5

Performance comparison of different methods.

Authors	Imaging modality/biomarkers	Source of data; (AD/CN)	Repetition; (cross validation)	Accuracy; (%)	Sensitivity; (%)	Specificity; (%)
Zhang et al., 2011 [20]	MRI	ADNI (51/52)	10 (10 folds)	86.2	86	86.3
Zhang et al., 2011 [20]	CSF	ADNI (51/52)	10 (10 folds)	82.1	81.9	82.3
Zhang et al., 2011 [20]	PET	ADNI (51/52)	10 (10 folds)	86.5	86.3	86.6
Zhang et al., 2011 [20]	MRI, PET, CSF	ADNI (51/52)	10 (10 folds)	93.2	93.0	93.3
Hinrichs et al., 2011 [31]	MRI + PET	ADNI (48/66)	30 (10 folds)	87.6	78.9	93.8
Hinrichs et al., 2011 [31]	MRI + PET + CSF + APOE + cognitive scores	ADNI (48/66)	30 (10 folds)	92.4	86.7	96.6
Magnin et al., 2009 [24]	MRI	Private (16/22)	5000 (75% training/25% testing)	94.5	91.5	96.6
Klöppel et al., 2008 [22]	MRI	(Group I) private (20/20)	Leave-one-out	95.0	95.0	95.0
Klöppel et al., 2008 [22]	MRI	(Group II) private (14/14)	Leave-one-out	92.9	100	85.7
Klöppel et al., 2008 [22]	MRI	(Group III) private (33/57)	Leave-one-out	81.1	60.6	93.0
Walhovd et al., 2010 [32]	MRI	ADNI (42/38)	N/A	82.5	81.6	83.3
Walhovd et al., 2010 [32]	MRI + CSF	ADNI (42/38)	N/A	88.8	86.8	90.5
Cuingnet et al., 2011* [12]	MRI	ADNI (162/137)	N/A (2 folds)	N/A	81.0	95.0
Proposed study	MRI + MMSE	Private  (129/60)	50 (5 folds)	92.3	88.2	94.2

∗This paper by Cuingnet et al. [12] compares ten methods and the best performance is shown here.