Table 9.

Comparison of the performance of different multi-modal classification algorithms

Algorithms	Subjects	Modalities	AD vs NC	MCI vs NC	MCI-C vs MCI-NC	Algorithm Description
MKL (Zhang et al., 2011)	51AD, 43MCI-C, 56MCI-NC, 52NC	MRI + PET +CSF	93.20	76.40	--	The classical multi-kernel learning (MKL) based algorithm
MTL (Jie et al., 2015)	51AD, 43MCI-C, 56MCI-NC, 52NC	MRI + PET +CSF	95.03	79.27	68.94	The multi-task learning (MTL) based algorithm
M-RBM (Suk et al., 2014)	93AD, 76MCI-C, 128 MCI-NC, 101 NC	MRI + PET	95.35	85.67	75.92	The pioneering multi-modal deep RBM (M-RBM) based feature learning algorithms
SAE (Liu et al., 2015b)	85AD, 67MCI-C, 102 MCI-NC, 77 NC	MRI + PET	91.35	90.42	--	The SAE-based multi-modal neuroimaging feature learning algorithm
SAE-MKL (Suk, 2013)	51AD, 43MCI-C, 56MCI-NC, 52NC	MRI + PET +CSF	98.80	90.70	83.30	The combination of SAE-based feature learning and MKL classification (SAE-MKL) algorithm
DW-S2MTL (Suk et al., 2016)	51AD, 43MCI-C, 56MCI-NC, 52NC	MRI + PET +CSF	95.09	78.77	73.04	The deep sparse multi-task learning based feature selection (DW-S2MTL) algorithm
Dropout-DL (Li et al., 2015)	51AD, 43MCI-C, 56MCI-NC, 52NC	MRI + PET +CSF	91.40	77.40	70.10	The dropout based robust multi-task deep learning (Dropout-DL) algorithm
SDSAE (Shi et al., 2017)	94AD, 121MCI, 123NC	Longitudinal MRI	91.95	83.72	--	The SDSAE-based feature learning algorithm
NGF (Tong et al., 2017)	37AD, 75MCI, 35NC	MRI + PET +CSF + Genetics	98.10	82.40	77.90	The nonlinear graph fusion (NGF) based algorithm
MM-SDPN-SVM (Shi et al., 2018)	51AD, 43MCI-C, 56MCI-NC, 52NC	MRI + PET	97.13	87.24	78.88	The multi-modal stacked deep polynomial networks and SVM