[1P0Y] Pipeline reproduction (SPM, FSL, ANTs - raw)

[bclenet]

Softwares

• SPM12
• FSL-6.0.0
• Advanced Normalisation Tools (ANTs) version 2.3.1
• ART tools within CONN for Connectivity Toolbox

Input data

raw data

Additional context

see description below

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).

---

NARPS team description : 1P0Y

General

• teamID : 1P0Y
• NV_collection_link : https://neurovault.org/collections/5649/
• results_comments : Hypothesis 1: We observed a significant medial frontal cluster, however this was more superior, localised more to dorsal anterior cingulate and medial superior frontal cortex. The lowest Z co-ordinate (MNI) for local peaks within this cluster was 30.

Hypthesis 2: As above, we observed a significant medial frontal cluster, however this was more superior, localised more to dorsal anterior cingulate and medial superior frontal cortex. The lowest Z-co-ordinate for local peaks within this cluster was 32.

Hypothesis 3 & 4: The implicit mask generated in group-level statistical analysis excluded significant portions of this area, hence we cannot be confident in our findings for these regions.

Hypothesis 5: A significant cluster of 288 voxels (peak MNI co-ordinates: -10, 60, -8) was observed. Anatomical labeling of the cluster using AAL2 indicated the majority of the cluster was located within bilateral medial orbital frontal gyrus and rectus.

Hypothesis 6: A significant cluster of 94 voxels (peak MNI co-ordinates: -10, 46, -12) was observed. Anatomical labeling of the cluster using AAL2 indicated the cluster was located within bilateral medial orbital frontal gyrus and rectus.

Hypothesis 7: Significant positive effect of loss was observed in 9 clusters, including bilateral Insula cortex, bilateral occipital clusters, supplementary motor area and a left parietal region. No signifcant clusters encompassing amygdala were observed.

Hypothesis 8: Significant positive effect of loss was observed in 16 clusters, including bilateral Insula cortex, bilateral occipital clusters, supplementary motor area and parietal regions. No signifcant clusters encompassing amygdala were observed.

Hypothesis 9: Four clusters showed greater positive response to loss in the equal range group compared to the equal indifference condition. Two clusters in right frontal regions and bilateral inferior parietal regions showed this group difference in positive parametric response to loss. No effect was observed in amygdala.

• preregistered : No

• link_preregistration_form : NA

• regions_definition : We used for Automated Anatomical Labeling Atlas (V2) to define masks for the regions of interest. We followed the description of Rolls et al. (2015) for the vmPFC, using the updated Automated Anatomical Labeling Atlas, including bilateral gyrus rectus, medial superior frontal gyrus, medial orbital frontal gyrus, medial orbital gyrus, anterior cingulate cortex (‘AAL2’; Rolls et al., 2015). AAL2 includes a labelled region comprising amygdala, which formed the mask for this regional hypothesis.

• softwares : SPM12 ,
  FSL-6.0.0 ,
  Advanced Normalisation Tools (ANTs) version 2.3.1, ART tools within CONN for Connectivity Toolbox

• general_comments : NA

Exclusions

• n_participants : 104
• exclusions_details : Where the percentage of volume outliers identified for scrubbing based on mean signal or framewise displacement exceeded 10% of total volumes, the participant was excluded from group-level analyses. This criteria excluded four particpants from analysis (sub-016, sub-030, sub-116, sub-120).

Preprocessing

• used_fmriprep_data : No
• preprocessing_order : Realignment of EPI time-series (FSL: mcflirt)
  T1 Brain Extraction (ANTs: antsBrainExtraction)
  T1 Segmentation (FSL: fast)
  Fieldmap created (FSL: fsl_prepare_fieldmap)
  Coregistration of EPI time-series to T1 and EPI distortion correction (FSL: epi_reg)
  Masking of non-brain voxels from EPI time-series (FSL: fslmaths)
  T1 normalisation to MNI template (ANTs: antsRegistration, antsApplyWarps)
  EPI time-series normalisation (ANTs: antsApplyWarps)
  EPI time-series smoothing (FSL: fslmaths)
• brain_extraction : T1 brains were extracted using 'antsBrainExtraction.sh' from the ANTs toolbox using the IXI brain template and probability mask
• segmentation : FSL fast was used to extract grey matter, white matter and CSF from the T1-weighted image. Additionally, the bias field and brain extracted bias corrected T1 were saved.
• slice_time_correction : No slice-timing correction was performed
• motion_correction : Motion correction was performed with reference to the single band reference image using 'mcflirt' from FSL.
• motion :
• gradient_distortion_correction : No gradient distortion correction was performed.
• intra_subject_coreg : EPIs were coregistered to T1 structural image by initially creating warp fields with 'epi_reg' from FSL. This process utilised brain-extracted, bias field corrected T1 image that was computed during brain extraction step in addition to the original T1 (both resampled to 2mm isotropic voxels using linear interpolation with ANTs 'ResampleImage'), a white matter image that was obtained during segmentation then binarised (threshold = 0.9) with 'fslmaths', a field map in radians computed using 'fsl_prepare_fieldmap', and both a the original magnitude field map image and a brain extracted version computed with the same aforementioned T1 brain extraction method. The warps were then applied to the EPI time-series using 'applywarps' from FSL using spline interpolation.
• distortion_correction : Susceptibility distortion correction of the EPI time-series was conducted as part of the 'epi_reg' routine described above.
• inter_subject_reg : Warp fields describing the transformation of the bias corrected T1 image to the 2mm MNI template space supplied with FSL was determined using the default transformation option ('s') for ANTs' non-linear registration tool 'antsRegistrationSyn.sh'. This tool includes rigid, affine and diffeomorphic symmetric registration processes. The warp fields generated during this process were then applied to T1 data with 'antsApplyTransforms'. Subsequently, these warp fields were applied to coregistered EPI data. Note the time-series were first split using 'fslsplit' into 3D volumes and the warps applied to individual images - we could not get the ANTs routine to work on a 4D dataset; after normalisation of the time-series, 'fslmerge' was used to concatenate the data into a 4D file for each run.
• intensity_correction : A bias field corrected T1 image was computed during brain extraction.
• intensity_normalization : First level modelling was performed using SPM12, hence each run was scaled such that the mean image had a mean intra-cerebral intensity of 100.
• noise_removal : We applied ICA-AROMA to EPI time-series after spatial smoothing using the 'non-aggressive' algorithm setting. We also concatenated motion-related noise regressors to be included as covariates in first level imodels, including the motion parameters computed by mcflirt, and their first temporal derivative that was computed using the central difference method.
• volume_censoring : We used ART tools (within the CONN for connectivity toolbox) to compute EPI volume outliers. An image was deemed an outlier if it was >3 standard deviations from the time-series mean (within run) or exhibited > 2mm framewise displacement. One regressor was created for each outlier with a '1' flagging the outlier and '0' identifiying each other time-series image.
• spatial_smoothing : Spatial smoothing was applied using a 5mm FWHM kernel (kernel divided by sigma where 2.3548) within fslmaths.
• preprocessing_comments : Data were always written using float datatype precision.

Analysis

• data_submitted_to_model : As detailed above, four participants with >10% of volumes exceeding criteria for outlier scrubbing were excluded from final analysis.
• spatial_region_modeled : Full brain
• independent_vars_first_level : At the first level of analysis, effects of interest (Gain and Loss effects) were assessed in separate event-related models. Each model included an 'All Trials' regressor that coded the onsets of all trials with 4 second durations, and incorporated three linear parametric modulation regressors. The first regressor modelled trial decision responses, either 'accept' (coded as '1s') and 'reject' (coded as '-1s') responses and the second parametric modulation regressor modelled 'expected value' (as defined in Canessa et al., 2013). In the 'Gain' model, the third regressor modelled 'Gain' amounts for each mixed gamble trial, whereas in the 'Loss' model, loss amounts were coded for each mixed gamble trial. Covariates of no interest included the motion-related noise and volume censoring regressors in both models. Slow signal drift was removed using 1/128 Hz high-pass filter, and serial correlations in the timeseries accounted for using an AR(1) model during (Restricted Maximum Likelihood) parameter estimation. Contrast images depicting the linear effect of Gain and Loss were computed using one sample t-tests on the gain and loss regressor for the respective models. Note that SPM12 orthogonalises parametric modulation regressors such that any shared variance between regressors is attributed to the first regressor modelled. Gain and Loss contrast images were entered into second level models as appropriate.
• RT_modeling : none
• movement_modeling : 0
• independent_vars_higher_level : Contrast images assessing the linear effect of Gain and Loss were entered into second level random effects models as implemented in SPM12. Each directional hypothesis was assessed using a one-sided T-test. For hypotheses 1-8, this comprised one-sample T-tests, with one model using the Gain contrast image for those in the equal indifference group (hypotheses 1 & 3), another the Gain contrast image for the equal range group (hypotheses 2 & 4), another the Loss contrast images for the those in the equal indifference group (hypotheses 5 & 7), and a model for the Loss contrast image for the equal range group (hypotheses 6 & 8). Similarly, hypothesis 9 was assessed by a entering the Loss contrast for the two groups into a two-sample T-test, with one-sided contrast used to assess equal range > equal indifference conditions.
• model_type : Mass Univariate
• model_settings : As above, SPM defaults: global approximate AR(1) (first level); Random effects, Ordinary Least Squares.
• inference_contrast_effect : One-sided contrasts were used to assess each hypothesis within the one-sample T-test (hypotheses 1-8) and two-sample T-tests (hypothesis 9).
• search_region : Regional hypotheses were masked using anatomical Regions of Interest, derived from the Automated Anatomical Labeling Atlas (version 2) as described above. Correction for multiple comparisons was conducted at the whole brain level.
• statistic_type : As implemented in SPM, cluster extent based thresholding was used, with a voxel-wise cluster-defining threshold of p<.001.
• pval_computation : Standard parametric inference
• multiple_testing_correction : Cluster-level FWE corrected p<.05 was used for all hypothesis testing, using Random Field Theory, as implemented in SPM.
• comments_analysis : NA

Categorized for analysis

• region_definition_vmpfc : atlas AAL
• region_definition_striatum : atlas AAL
• region_definition_amygdala : atlas AAL
• analysis_SW : SPM
• analysis_SW_with_version : SPM12
• smoothing_coef : 5
• testing : parametric
• testing_thresh : p<0.001
• correction_method : GRTFWE cluster
• correction_thresh_ : p<0.05

Derived

• n_participants : 104
• excluded_participants : 016, 030, 116, 120
• func_fwhm : 5
• con_fwhm :

Comments

• excluded_from_narps_analysis : No
• exclusion_comment : N/A
• reproducibility : 2
• reproducibility_comment :