[4SZ2] Pipeline reproduction (FSL, deriv)

[michellewang]

Softwares

FSL 6.0

Input data

derivatives (fMRIprep)

Additional context

Preprocessing:

• preprocessing_order: "First level analysis with FSL, including temporal filtering and data whitening. Subsequently, the second level analysis was performed on the COPE maps."
• brain_extraction: "We used the fmriprep provided brain masks."
• segmentation: "No segmentations was used."
• slice_time_correction: "Provided by fmriprep."
• motion_correction: "Provided by fmriprep."
• gradient_distortion_correction: "Provided by fmriprep."
• intra_subject_coreg: "Provided by fmriprep."
• distortion_correction: "Provided by fmriprep."
• inter_subject_reg: "Provided by fmriprep."
• intensity_correction: "Not performed"
• intensity_normalization: "Performed by FSL."
• noise_removal: "Not accounted for."
• volume_censoring: "The "robust" option of FSL was tested, but did not lead to noticeable differences."
• spatial_smoothing: "Performed by FSL, 5mm FWHM."
• preprocessing_comments: "We used 100Hz temporal filtering."

Analysis:

• data_submitted_to_model: "Each run of each subject was analyzed independently using all the 453 volumes."
• spatial_region_modeled: Full brain
• independent_vars_first_level: "The values of GAIN and LOSS were used as independent columns of the regression matrix. These were convolved with the canonical HRF. The time derivatives of the two columns were also included in the regression."
• RT_modeling: None
• movement_modeling: 0
• independent_vars_higher_level: "For the second level analysis, 2 columns of the regression matrix were used to separate between equalRange and equalIndifference. Each run of each subject was treated as independent. Further, age and sex were added as nuisance factors."
• model_type: Mass univariate
• model_settings: "We used FSL FEAT with data whitening. The regressors were temporally filtered with the same settings used for the data. The estimation was performed with FLAME 1."
• inference_contrast_effect: "Hypothesis 1-8 resulted from second level contrasts selecting a single column, such as [1 0],[0 1]. These were applied to positive and negative effects separately. Hypothesis 9 was tested as [-1 1] (being equalRange the second column) for both positive and negative LOSS effects."
• search_region: Whole brain.
• statistic_type: Default TFCE was used.
• pval_computation: Standard FEAT pipeline.
• multiple_testing_correction: Cluster p-threshold 0.05.

List of tasks

Please tick the boxes below once the corresponding task is finished. 👍

• 👌 A maintainer of the project approved the issue, by assigning a 🏁status: ready for dev label to it.
• 🌳 Create a branch on your fork to start the reproduction.
• 🌅 Create a file team_{team_id}.py inside the narps_open/pipelines/ directory. You can use a file inside narps_open/pipelines/templates as a template if needed.
• 📥 Create a pull request as soon as you completed the previous task.
• 🧠 Write the code for the pipeline, using Nipype and the file architecture described in docs/pipelines.md.
• 📘 Make sure your code is documented enough.
• 🐍 Make sure your code is explicit and conforms with PEP8.
• 🔬 Create tests for your pipeline. You can use files in tests/pipelines/test_team_* as examples.
• 🔬 Make sure your code passes all the tests you created (see docs/testing.md).

[bclenet]

Hi @cmaumet,

Here are quick questions about the analysis. (Full description above)

independent_vars_first_level : The values of GAIN and LOSS were used as independent columns of the regression matrix. These were convolved with the canonical HRF. The time derivatives of the two columns were also included in the regression.

Does this mean we use the canonical HRF + derivatives as basis ?

independent_vars_higher_level : For the second level analysis, 2 columns of the regression matrix were used to separate between equalRange and equalIndifference. Each run of each subject was treated as independent. Further, age and sex were added as nuisance factors.

Here, I assume second level analysis refers to group level. Is there intermediate subject level ?

Thanks !

[cmaumet]

Does this mean we use the canonical HRF + derivatives as basis ?

Yes exactly!

Here, I assume second level analysis refers to group level. Is there intermediate subject level ?

Yes that is right second level = group. From the "Each run of each subject was treated as independent", I think it means that there is no subject-level, i.e. run-level and then group-level (as if each run was a separate subject).

[bclenet]

Results with 108 subjects, commit 3ff9cf4
[0.20, 0.30, 0.20, 0.30, -0.41, -0.35, 0.41, 0.35, -0.11]