C. Atlas Preparation and FC Computation

1. Atlas Preparation

This project uses the 400 ROI version of the Schaefer Atlas sorted according to the 7 Yeo Networks. This atlas is publicly available here. Becuase this atlas does not include subcortical regions, we rely on the AAL atlas to obtain regions of interest for subcortical structures.

The preparation of this atlas for this project takes place in notebook S07_PrepareAtlases_400. Here we provide some instructions on how to proceed, but for additional details on atlas preparation please check that notebook.

To start, please do as follows:

1. Create a local folder for brain parcellations: (e.g., /data/SFIMJGC_Introspec/2023_fc_intrsopection/atlases)

2. Assign the full path of that folder to variable ATLASES_DIR in basics.py.

3. Assign the name of the atlas (Schaefer2018_200Parcels_7Networks) to the ATLAS_NAME variable in basics.py.

   NOTE: Unless you are using a different version of the Schaefer Atlas, you can skip this step.

4. Create a sub-folder for the 400 Schaefer Atlas:

cd ${ATLASES_DIR}
mkdir ${ATLAS_NAME}

5. Copy the following files from their original location in CBIG repo (link above) to your local Schaefer Atlas folder:

• Schaefer2018_400Parcels_7Networks_order_FSLMNI152_2mm.nii.gz: 400 ROI Atlas / 7 Yeo Networks in MNI space on a 2mm grid.

• Schaefer2018_400Parcels_7Networks_order.txt: list of all ROIs, their network membership and color code (RGB).

• Schaefer2018_400Parcels_7Networks_order.lut: same as above, but coded for FSL eyes.

• Schaefer2018_400Parcels_7Networks_order_FSLMNI152_2mm.Centroid_RAS.csv: information about each ROI centroid coordinates.

6. Create a folder for the aal2 atlas

cd ${ATLASES_DIR}
mkdir aal2

7. Download the AAL v2 atlas from here

8. Unzip the contents of the downloaded file into ${ATLASES_DIR}/aal2. Only aal2.nii.gz and aal2.nii.txt are needed.

Once you complete steps 1 - 8, go to S07_PrepareAtlases_400 and continue there. Some of the instructions shown above are also in the notebook. No need to perform the steps twice.

2. Extraction of ROI representative timeseries

At this point we have our data fully pre-processed and an atlas ready to use. The next step is to extract represenative timeseries per ROI/scan. This is accomplished via the S08_Extract_ROI_ts notebook, which relies on AFNI program 3dNetCorr to do so.

3. Exploration of rs-fMRI Functional Connectivity matrices

Once FC matrices for all scans were available, we spent some time looking at them in the following way:

1. We computed the average FC matrix for the whole sample. For this, we first Fisher transformed each individual matrix prior to averaging. Once the average was available, we did the inverse transformation to obtain the FC matrix representative of the whole sample.

2. We explored all the matrices either in the form of a video that shows them one after the other or by plotting them in the form of a carpet plot. We did this with different sorting schemes, including: correlation to the mean, covariance with the mean, cosine distance to the mean, correlation distance to the mean, and euclidean distance to the mean.

The notebook that contains the code used for these explorative analyses is S09_ExploreFCmatrices_Schaefer_AAL2. This notebook will not only plot figures on the notebook itself but also launch two different dashboards:

a) Carpet Plot dashboard: allows us to explore the whole sample in this form.

Carpet Plot Dashboard

b) Video Player dashboard: allows us to explore the whole sample by plotting each scan FC matrix in succession.

Video Player Dashboard

Both dashboards contain controls for selecting sorting the sorting scheme.