06 - first level GLM

Several approaches to run a first level GLM using nilearn

After preprocessing your data, probably with fmriprep, it´s time to do your first level GLM (= general linear model).

In this step, you will compute mean contrast-estimates for predefined contrasts for each subject individually. These estimations will be used later in the second level Analysis, where you extract relevant (for your research question/hypothesis) values.

For all cases described below, there are generally three things you have to change before running the cells in the jupyter notebook (or the python script):

1. Change the directories

2. Indicate the task, the number of sessions and the subject

3. Implement your contrasts

1 Surface-based Analysis

To compute a first-level GLM you can adapt the following jupyter notebook: /shared/nilearn_ffx.ipynb

You can either run each cell separately (“Play”-Button at the top) after implementing the changes (if changes are needed), or first implement all changes and then run all cells automaticall (“Double-Play”-Button at the top)

1. In the second cell (indicated with [2]: at the left), change:

1. data_dir

2. fmriprep_dir

3. task_label (this refers to how your files are named! \(\to\) in your BIDS-valid naming of your file, what´s behind “task-” \(\to\) For sub-12002_ses-1_task-loc_run-1_space-T1w_desc-preproc_bold.nii.gz the corresponding task_label would be loc \(\to\) task_label = "loc")

4. sub_labels

5. After the comment # Loop over sessions: indicate the number of sessions, by “counting” from 1 to the number of sessions you did (if you did 4 sessions, the line should read for ses in [1, 2, 3, 4])

2. In the third cell (indicated with [3]: at the left), change the number of sessions similar to step 1e)

3. In the sixth cell (indicated with [6]: at the left):

1. Define your constrasts-dictionairy, by defining a name for each contrast (= key) and indicating how it is computed (= value)

2. for indicating how a contrast is computed, you can access the columns of the design matrix (that should be printed above this cell). Keep in mind, that for basic_contrasts[] you indicate the actual name of the column, not the index!

3. you can print the resulting contrasts matrices, but you don´t have to

4. In the seventh cell (indicated with [7]: at the left):

1. define empty lists for each of your contrasts

2. change the respective names of the lists before the .append statement

3. indicate the correct contrast names in the parameters of the .compute_contrast function

5. In the eighth cell (indicated with [8]: at the left): change contrasts_dict to match with your lists

6. In the second part of the eighth cell (somehow the enumeration is messed up there): comment out (using #) the line starting with data_dir = (since you should have writing access in your own directories and you already defined data_dir in the beginning of the script)

2 Volume-based Analysis

There is also a script for Volume-based Analysis