As in earlier work32 and in keeping with recommendations by Calhoun and de Lacy,33 75 components were extracted. Individual components were reviewed and components that (1) had low power in the canonical 0.01 to 0.15 Hz frequency spectrum, (2) had signal primarily in white matter or cerebrospinal fluid, or (3) had motion-related “ringing” artifact present on the images were discarded. Specifically, all components were reviewed by three of the authors (A.E., A.W., and J.V.). Ratings were made on a 3-point scale (1 = retain; 2 = consider retaining, 3 = reject). Only components that had at least two retain votes and no reject votes were ultimately included. This approach resulted in 29 retained components (Figure 1). Components were labeled based on anatomical regions involved and correspondence to other ICA-based network results.34
