46. Resting‐state fMRI analysis

These are procedures for single-subject resting-state fMRI analysis using 'seed-based correlation'. Namely, the time series at a given "seed" brain location is used to correlate with the fMRI time series at other brain locations. The distribution of brain regions showing significant correlation to the chosen seed consists of a resting-state network.

The typical choice of seeds are medial prefrontal cortex, bilateral superior parietal labules, and precuneus. These are sites considered for a "default-mode network".

Here we chose the hippocampus as the seed region for its functional role in memory.

Other non 'seed-based correlation' methods for identifying resting-state brain networks, such as principal component analysis or independent component analysis, are covered elsewhere.

The scripts were tested at /space_lin2/fhlinl/eegmri_memory/s006/resting_analysis on January 7 2024.

1. Preparation

fMRI data should be processed according to the description. In this pre-processing, fMRI data should be co-registered to structural MRI with FreeSurfer reconstruction. The procedure here illustrates the co-registration in details.

2. Collect confounds

The confound here indicates the time series not related to the seed dynamics. Here we consider motion time series, including rotation and translation, and the signal fluctuations at the ventricles and the white matter as confound.

The description here includes two scripts to get the these confounds.

3. Seed time series

FreeSurfer automatically created cortical surface "parcellation" (*.annot) and sub-cortical segmentation (.aseg) based on anatomical MRI. Here we spatially transform these information from the anatomical space (defined by structural MRI, MPRAGE, slices) to function space (defined by fMRI slices). Then we extract the time series averaged across image voxels in the hippocampus.

This is done by this script.

4. Perform correlational analysis

Confounds, including head motion and time series at ventricles and the white matter, were first removed from the whole brain MRI data. Then correlation coefficients were calculated at each brain location using the seed defined above.

This is done by this script.

5. Render results over the brain

Now we visualize the functional connectivity to the hippocampus using this script.

---

Here are procedures to repeat the above analysis for subject s026 in project seeg (tested on March 21 2024)

3b. Seed time series

This script extracts the time series at the hippomcampus as the seed time series in the functional connectivity analysis using anatomical parcellation created by FreeSurfer. Only one run (run 032) was used.

4b. Perform correlational analysis

This script uses the seed time series to perform functional connectivity analysis by calculating the correlation coefficients at all brain locations with respect to the "seed" (the hippocampus).

5b. Render results over the brain

This script shows the functional connectivity map with the hippocampus as the seed. Note that the map is at the subject's native volume space.

6. Morph results from native volume to native and atlas surfaces

This script morphs the functional connectivity map in subject's native volume to subject's native surface space and atlas (fsaverage) surface space in the STC format. Values over the atlas surface allows for group-level analysis.

7. Render morphed results using values on surfaces

###Show the maps on subject's native space using STC files.

• Show s026 left hemisphere brain

etc_render_fsbrain('subject','s026');

• Load hippo_fconn_native_vol_aseg_032124_gavg_hippo_left-anat-lh.stc (morphed functional connectivity maps in STC).

Hint: Press f to load STC files as overlays.

• Change the threshold to [1.5 and 4]

Hint: Press d to change the overlay threshold.

Show the maps on atlas space using STC files.

• Show fsaverage left hemisphere brain

etc_render_fsbrain;

• Load s026_2_fsaverage_hippo_fconn_032124-lh.stc (morphed functional connectivity maps in STC).

• Change the threshold to [1.5 and 4]