Neuroimaging Tutorial with CONN
CONN is an open source SPM/Matlab based imaging software for the analysis of functional connectivity in fMRI during task and resting state.
In this first post, I will illustrate the required steps involved in importing and pre-processing functional and structural resting-state data acquired from 25 subjects.
What you need to start
Launch Matlab and add CONN and SPM into your path, using File > Set Path > Add Folder… or typing:
>> addpath C:\Users\login\Documents\MATLAB\conn
Then you can launch CONN by typing in the command window:
Click on New to initialize a new project, choose a name and save it.
Here we only need to modify the Number of Subjects to 25, since in our dataset we have 25 scans. Then we are ready to import our NYU dataset.
In the picture above, on the right, you can see that we are inside the subfolder “anat”. mprage_anonymized.nii.gz is the name of the first structural scan. Select it and some information regarding the brain regions and voxels position will display when you move the cursor over the image. In order to import the 25 scans together, select your NYU folder. Type the filename (mprage_anonymized.nii.gz) in the same space where all the different file formats are reported and click on Find. Subsequently highlight your 25 subjects on the left and then highlight your 25 scans on the right and click on Select:
Repeat the same process to import the functional data:
After importing both structural and functional scans, we can start with Pre-Processing. As you can see below, few data pipeline are already present. For this tutorial we will use the first one, which involves realignment, slice-timing correction, outlier identification, normalization, segmentation and smoothing.
As we start this pipeline, CONN will ask four questions. In the first one we need to choose the slice order, and we will select “Interleaved Siemens”. Indeed interleaved is particular on Siemens machines as this mode always acquire in ascending fashion. The second question asks which threshold we want apply in the outlier detection, and 95th percentile is the recommended one:
Then, CONN will ask how we want to reslice the data after normalization, and we can type 2mm:
Finally, we define a value of 8mm for our spatial smoothing:
At this point the Pre-Processing will start and will take a while, approximately 5 mins for each subjects!