Pipeline for generating functional conncectivity and structural connectivity based on fmriprep and qsiprep


Summary of what happened:

I’m a computer science major with little knowledge of neuroscience. And I want to do some research on brain network. After being tormented by the dizzying toolchain and poor expertise in the field of neuroscience, I decided to focus my energy on fmriprep, nilearn and qsiprep. Here are my questions:

  1. After preprocessed with fmriprep, I got images in MNI152NLin2009cAsym space (Please remind me if I use inappropriate words or phrases), and I want to use the command atlas_aparc = datasets.fetch_atlas_destrieux_2009() to access the atlas. So the question is: does the space alignment matter during network construction? If so, how can I know which space is the atlas provided by nilearn in?

  2. If I want to use some atlas not avaliable in nilearn (e.g. brainnetome246) or qsiprep (e.g. destrieux_2009), how can I realize? I notice that in document of Qsiprep,we can realize by

"my_custom_atlas": {
    "file": "file_in_this_directory.nii.gz",
    "node_names": ["Region1_L", "Region1_R" ... "RegionN_R"],
    "node_ids": [1, 2, ..., N]

But how can I get "file_in_this_directory.nii.gz"? Are they avaliable in freesurfer output?

  1. Are there any recommended systematic textbooks or tutorials like Andy’s Brain Book suitable for beginners? I’m still struggling with concepts in neuro-image and neuro-science.


Hi @taka2223, maybe I can answer your questions partially for the nilearn side.

  1. Nilearn assumes standard MNI152 space so all the volumetric data we provide access to is in MNI space. It expects users to have their data in this space to successfully use the default functionality. There are functions to resample the data/atlas to match their resolutions and shapes but you can’t register an image from one space (e.g. native subject space) to another (e.g. standard) using nilearn. Someone correct me if I’m wrong, but MNI152NLin2009cAsym is the default fmriprerp output and should be compatible with nilearn (it’s in MNI152 space as the name suggests and the part “NLin2009cAsym” refers to nonlinear registration and asymmetric version of the template. This is one of the most common volumetric templates as far as I know).

  2. To use your own atlas or data in general from nifti files you can just load it with nilearn.image.load_img

Let me know if anything is not clear.

1 Like