Inria-Empenn / narps_open_pipelines

A codebase reproducing the 70 pipelines of the NARPS study (Botvinik-Nezer et al., 2020) shared as an open resource for the community.
MIT License
9 stars 22 forks source link

[T54A] Pipeline reproduction (FSL - deriv) #76

Open bclenet opened 10 months ago

bclenet commented 10 months ago

Softwares

FSL v 5.0.9; RRID:SCR_002823 , fmriprep v.1.1.4

Input data

derivatives (fMRIprep)

Additional context

This pipeline is (almost) reproduced, only refactoring should be needed.

List of tasks

Please tick the boxes below once the corresponding task is finished. :+1:

bclenet commented 5 months ago

Here are the correlation results for T54A. Improvement is needed.

Team Number of subjects Test status Correlation values
T54A 108 failure [0.83, 0.72, 0.83, 0.72, 0.70, 0.70, 0.70, 0.70, -0.01]
bclenet commented 2 months ago

It looks like the regressors of interest are not orthogonalized in the code, although mentioned in the description (section independent_vars_first_level):

  • Gain: Modeled for trials with a valid response (Not NoResp), [...], orthoganlized with respect to the constant factor.
  • Loss: Modeled for trials with a valid response (Not NoResp), [...], orthoganalized with respect to the constant factor.
  • Difficulty: Modeled for trials with a valid response (Not NoResp), [...], orthogonalized with respect to the constant factor.

This can be done using the orthogonalization parameter in FSL's Level1Design.