Figure 3.

Hand-identified resting state networks found with DICA and ICA over 5 re-runs. In each row, the results for 5 re-runs are shown, where the L = 14 IC maps were inspected to find the visual (top 2 rows) and auditory (bottom 2 rows) networks. DICA with random ICA initialization was used (DICA-RI). For the visual networks, the 5 re-runs are fairly consistent, though DICA appears to exhibit better consistency than ICA. For auditory, both exhibit variation with perhaps DICA having greater variability. In particular, considering the auditory network on the 3rd and 4th runs of DICA, a user might make quite different conclusions about the involvement of this subject’s superior temporal gyrus in auditory processing depending on which run’s results they saw.