48. Group‐level fMRI analysis
This is the 2nd-level group analysis for modeling the fMRI time series using General Linear Model (GLM). See this page for the 1-st level within-subject analysis. The scripts were tested at /space_lin2/fhlin/seeg/average/fmri on March 13 2024.
NOTE: At Sunnybrook, the data and scripts are at /space_lin2/fhlin/seeg.
Functional MRI data from five participants (s025, s026, s031, s032, s034) should be analyzed first with the following information:
/space_lin2/fhlin/seeg/s025/fmri_data/dicom
/space_lin2/fhlin/seeg/s026/fmri_data/dicom
/space_lin2/fhlin/seeg/s031/fmri_data/dicom
/space_lin2/fhlin/seeg/s032/fmri_data/dicom
/space_lin2/fhlin/seeg/s034/fmri_data/dicom
/space_lin2/fhlin/seeg/subjects/s025
/space_lin2/fhlin/seeg/subjects/s026
/space_lin2/fhlin/seeg/subjects/s031
/space_lin2/fhlin/seeg/subjects/s032
/space_lin2/fhlin/seeg/subjects/s034
/space_lin2/fhlin/seeg/s025/fmri_analysis/SOA_s025.mat
/space_lin2/fhlin/seeg/s026/fmri_analysis/SOA_s026.mat
/space_lin2/fhlin/seeg/s031/fmri_analysis/SOA_s031.mat
/space_lin2/fhlin/seeg/s032/fmri_analysis/SOA_s032.mat
/space_lin2/fhlin/seeg/s034/fmri_analysis/SOA_s034.mat
Check the t-statistics maps of each condition for each individual. Here are images for condition "A", "V", and "AV" for subject s034. This script is useful in generating a composite figures of the STC files for each condition and each individual. Note that the t-statistics were calculated from the data transformed to fsaverage and results were rendered on the inflated cortical surfaces. It is important to have individual's results calculated on the same space (MNI305 fsaverage subject in this case) because the subsequent group-level analysis needs data from individuals are spatially aligned.
subject s034 condition "A"
subject s034 condition "V"
subject s034 condition "AV"
Visualizing the activation in the same condition but across individuals is good to understand the inter-subject variability. Here are maps of for condition "AV" for all five subjects.
subject s025 condition "AV"
subject s026 condition "AV"
subject s031 condition "AV"
subject s032 condition "AV"
subject s034 condition "AV"
Use individual's beta maps, not t-statistics, to calculate the group-level activation. Use this script to calculate and show results. Here we calculate Z-scores for each condition to summarize the activities across individuals.
condition "A"
condition "V"
condition "AV"
In previous visualization, we used an arbitrary threshold Z=3 as the threshold to show activated brain areas in the group analysis. More correct way is to consider the inflation of Type-I error (alpha) in searching for brain activation across the whole brain and correct such multiple comparisons. A common approach is to determine the statistics threshold using False Discovery Rate.
Here is the script that use the FDR corrected threshold to show activated brain areas with adjusted p values of 0.05 and 0.01.
condition "A" (threshold between Z=2.5 and Z=3.2 for corrected p=0.05 and p=0.01, respectively)
condition "V" (threshold between Z=3.0 and Z=3.7 for corrected p=0.05 and p=0.01, respectively)
condition "AV" (threshold between Z=2.6 and Z=3.3 for corrected p=0.05 and p=0.01, respectively)
