Dear Alfonso,

This time I am trying to replicate, as closely as possible, a previous study. In this work, they use the Power atlas to estimate connectivity between networks. Ultimately, they obtain a single connectivity measure between the somatosensory network of the hand and the cerebellum and the rest of the networks (e.g., Hand-DAN, Hand-DMN, etc). Each network is represented by a variable number of ROIs derived from the Power et al. 2011 atlas.

Two approaches come to mind for this. One would be to introduce the Power et al. 2011 atlas, extract the connectivity values from the matrix generated by CONN "resultsROI_condition," and for example average the connectivity for each node of the Hand network with each of the nodes in the DAN network. The second, which would be easier for me, would be to create a seed for each network composed by the spheres that conforme the individual networks, so that I obtain a "seed" for each network made up of the spheres that constitute that network, thus extracting a single connectivity value for each subject and inter-network connection (e.g. Hand-DAN).

Are these approaches analogous? The procedure I want to imitate is the following:

"Time-series from each ROI were used to generate correlation matrices among all ROIs, and were then z-transformed to generate normalized correlation matrices for each participant. The following networks were the focus of the present study: somato-motor Hand (Hand; 30 ROIs), Default Mode (DMN; 58 ROIs), Salience (SAL; 18 ROIs), Cingulo-Opercular (CO; 14 ROIs), Fronto-parietal (FP; 25 ROIs), Dorsal Attention (DAN; 11 ROIs), Cerebellar (Cer; 4 ROIs), and Ventral Attention (VAN; 9 ROIs) (see Supplementary Material and Varangis et al., 2019 for further details and graphic illustration of the spatial distribution of these investigated networks of interest). Inter-network connectivity was estimated from the Hand and Cerebellar networks to the six primary cognitive networks (DMN, SAL, CO, FP, DAN, VAN) by computing the average correlation from the Hand and Cerebellar networks to each of the cognitive networks. Thus, there were 12 primary connectivity measures included in the present study: Hand-DMN, Hand-SAL, Hand-CO, Hand-FP, Hand-DAN, Hand-VAN, Cer-DMN, Cer-SAL, Cer-CO, Cer-FP, Cer-DAN, and Cer-VAN." DOI : https://doi.org/10.3389/fnagi.2022.956744

Additionally, the reference study employed the following pipeline: "The preprocessing pipeline included slice-timing and motion correction performed in FSL, calculation of frame-wise displacement, replacement for contaminated volumes (Carp, 2013), band-pass filtering using fslmaths, and residualization of the processed data with respect to FWD, root mean square difference of the BOLD signal, left and right hemisphere white matter, and lateral ventricular signals (Birn et al., 2006)."

If I want to imitate this pipeline as closely as possible, would it be correct to calculate the frame-wise displacement using CONN, include it in the list of confounds for denoising, and remove the realignment parameters from this list?

Thank you in advance for your guidance.

Sorry for such an extensive question, but I'm feeling a bit lost and would greatly appreciate your help.

Benxa