Table 7.

Comparison with the previous studies of AD vs. NC classification on ADNI dataset. The boldface denotes the best performance for each metric. (GM: Gray Matter; SVM: Support Vector Machine; CT: Cortical Thickness; PCA: Principal Component Analysis; LDA: Linear Discriminant Analysis; ROI: Region Of Interest; QDA: Quadratic Discriminant Analysis; RLR: Regularized Linear Regression; SAE: Stacked Auto-Encoder).

Method	Feature Type	Classifier	Subjects (AD/NC)	Accuracy (%)	Sensitivity (%)	Specificity (%)
Zhang and Shen (2012)	GM volumes	SVM	45/50	84.80	–	–
Cho et al. (2012)	CT	PCA+LDA	128/160	–	82.00	93.00
Coupé et al. (2012)	HP/EC volumes	QDA	60/60	90.00	88.00	92.00
Liu et al. (2012)	GM voxels	Ensemble SRC	198/229	90.80	86.32	94.76
Casanova et al. (2013)	GM voxels	RLR	171/188	87.10	84.30	88.90
Eskildsen et al. (2013)	ROI CT	LDA	194/226	84.50	79.40	88.90
Schmitter et al. (2015)	10 Volumes	SVM	221/276	–	86.00	91.00
Suk et al. (2015a)	GM volumes+SAE	SVM	51/52	88.20	–	–
Proposed	GM volumes	JLLR+DeepESM	186/226	91.02	92.72	89.94