Table 3.

Smoothing kernel sizes (column 2), subject-level (column 3) and group-level (column 4) analysis methods used in simulation and the corresponding AUCs (column 5) of each method when analyzing simulated data with a noise fraction of 0.85 (SNR = 0.18).

Abbreviations	Subject-level smoothing kernel size (FWHM)	Subject-level analysis method	Group-level analysis method	Area under the ROC curve (integrating over FPR ∈ [0,0.1])
SV+SV	0 mm	Univariate	Univariate	0.0712
SV+cCCA	0 mm	Univariate	Multivariate Method	0.0790
cCCA+SV	0 mm	Sum-cCCA	Univariate	0.0798
cCCA+cCCA	0 mm	Sum-cCCA	Multivariate Method	0.0849
SVGS2+SV	2 mm	Univariate	Univariate	0.0745
SVGS4+SV	4 mm	Univariate	Univariate	0.0748
SVGS6+SV	6 mm	Univariate	Univariate	0.0681
SV+Fisher	0 mm	Univariate	Fisher’s data fusion	0.0227