Table 2.

Number of single-subject ICs as proposed by MELODIC and number of potential RSNs (i.e., ICs that were not rated as obvious artifacts by visual inspection or the automated filter).

		Training set	Test set	Difference between sets
$N_n^{\text{src}}$	Range	35–48	29–84	pU = 0.57
	M	42.3	45.4
	SD	6.0	9.1
$N_n^{\text{RSN}}$ (visual insp.)	Range	11–17	3–30	pU = 0.77
	M	13.2	13.9
	SD	2.1	4.7
$N_n^{\text{RSN}}$ (filter)	Range	8–24	5–26	pU = 0.28
	M	13.3	15.6
	SD	5.8	5.5
Difference visual vs. filter		pU = 0.78	pU = 0.18

Test for equal medians: Mann-Whitney-Wilcoxon U-test.