Group_level analysis (statistics) of output after xcp_d processing

1

Hi,everyone.
I would like to second-level analysis for xcp_d output of surface-based func data from ALFF,ReHo and FC and can’t find appropriate tools.I try to use brainstat based on Python,however,fslr91k seemly doesn’t exist in the brainstat’s surface templates,which are only fslr32k,recommrnding to use nilearn or others? Simultaneously,I don’t konw if stat-alff,stat-reho and stat-pearsoncorrelation have been transformed to z-scores,if it not,does it need to be transformed before statistics analysis?
Any helpful proposals would be appreciated.

2

If brainstat’s fslr32k is a CIFTI, then it’s probably the same as the space XCP-D used.

None of those outputs have been transformed to z scores. ALFF and Reho wouldn’t typically be z-scored before analysis, but the Pearson correlations should be.

3

Dear tsalo
Thank for your prompt reply.
I failed to find standard fslr91k template in xcp_d source code on the github website.
Where can I find the standard fslr91k template and mask?
Best,
fhao

4

Hello everyone,

I have processed my resting-state fMRI data using XCP-D, and now I want to perform group-level statistical analysis on the surface-based outputs (ALFF, ReHo, and FC).

The output is in fsLR32k space, and I’m especially interested in identifying statistical differences between groups.

My questions:

  1. What are the recommended tools for surface-based group analysis of ALFF, ReHo, and FC maps?
  2. Should the Pearson correlation FC maps be Fisher z-transformed before analysis?
  3. Are there any tutorials or pipelines available for group-level analysis using BrainStat or PALM with these outputs?

Thank you in advance for your help!

5

Dear Oumayma
First,I want to ask you how to achieve xcp_d output in fsLR32k space,because I don’t find the 32k parameter in xcp_d official website,while the fmriPrep’s pre-processing parameters seems to be only 91K and 164K.
1.I may recommend to use VertexwiseR package based on R or Brainstat,for nilearn,it likely don’t support Cifti format.
2.The pearson correlation should be

3.The tutorials as followsVe
VertexwiseR:CogBrainHealthLab/VertexWiseR: VertexWiseR R package : a suite of tools to extract, manipulate, analyze, and visualize cortical surface and hippocampal vertex-wise data
Brainstat: BrainStat and Python Index — BrainStat 0.4.2 documentation

1 Like
6

Hi @fhaos123

Thank you so much for the clarification on fsLR32k — it makes much more sense now!

How I got it, by trying to downsample using HCP’s wb_command, but I’m not entirely sure if I’m doing it correctly. I need more details about the correct pipeline and understand it

Also, thanks a lot for recommending VertexwiseR and BrainStat — I’ll definitely give them a try

Thank u :pray: