2.3. Denoising

Denoising was performed using the CONN toolbox. As for preprocessing, we followed the same denoising described in our previous publications (Luppi et al., 2020, 2019): To reduce noise due to cardiac and motion artifacts, which are known to impact functional connectivity and network analyses (Power et al., 2012; van Dijk et al., 2012), we applied the anatomical CompCor method of denoising the functional data (Behzadi Y et al., 2007), also implemented within the CONN toolbox. The step of global signal regression (GSR) has received substantial attention in the literature (Andellini et al., 2015; Lydon-Staley et al., 2019; Power et al., 2014). Here, we chose to avoid GSR in favour of the aCompCor denoising procedure, in line with previous studies (Carhart-Harris et al., 2016; Luppi et al., 2019). Additionally, here one of our variables of interest is the proportion of anticorrelations between brain regions across different states; however, GSR mathematically mandates that approximately 50% of correlations between regions will be negative (Braun et al., 2012), thereby removing any potentially meaningful differences. The cartographic profile method employed here to identify integrated and segregated sub-states of dynamic functional connectivity (see below) has also been shown to be robust to the use of GSR (Shine et al., 2016).

The aCompCor method involves regressing out of the functional data the following confounding effects: the first five principal components attributable to each individual's white matter signal, and the first five components attributable to individual cerebrospinal fluid (CSF) signal (both computed by applying a one-voxel binary erosion step to the masks of voxels with values above 50% in the corresponding posterior probability maps (Whitfield-Gabrieli and Nieto-Castanon, 2012)); six subject-specific realignment parameters (three translations and three rotations) as well as their first-order temporal derivatives; scrubbing of the outlying scans previously identified by art during preprocessing (Placebo: mean = 1.16 ± 1.48% of volumes; LSD: mean = 1.63 ± 2.45% of volumes; max number of scrubbed volumes per scan: 7.4%); and main effect of scanning session. In CONN, confounding factors are removed separately for each voxel and for each subject and session, using Ordinary Least Squares (OLS) regression to project each BOLD signal timeseries to the sub-space orthogonal to all potential confounding effects (Whitfield-Gabrieli and Nieto-Castanon, 2012). Finally, linear detrending was also applied, and the subject-specific denoised BOLD signal timeseries were band-pass filtered to eliminate both low-frequency drift effects and high-frequency noise, thus retaining frequencies between 0.008 and 0.09 Hz.