incf-nidash/nidmresults-examples

A set of test data to generate NIDM-Results examples

fMRI Results comparison library

In this repository we share the results from fMRI analyses carried out in SPM, FSL and AFNI. Each of the directories contains the output from results analysed using a certain variant available in the respective software package, and a full overview of the testing variants we have applied can be found here.

Citation: Pauli R, Bowring A, Reynolds R, Chen G, Nichols TE and Maumet C (2016) Exploring fMRI Results Space: 31 Variants of an fMRI Analysis in AFNI, FSL, and SPM. Front. Neuroinform. 10:24. doi: 10.3389/fninf.2016.00024

nidmresults-examples's Issues

Contrast masking needs more than 1 contrast

Hi @AlexBowring! Starting to run the FSL test examples... I have noticed that fsl_contrast_mask only has one contrast defined while what we would really need is two contrasts and then mask contrast 1 with contrast 2. Could you have a look into this? Thanks!

Software version

Use the latest release for each software package (SPM, FSL, AFNI)

Add motion covariates in all analyses

License

We need to add a licence. I would say Public Domain Dedication and License (latest version) as this is the licence used by openfmri (cf. https://openfmri.org/faq/#licensed)

cluster size in resels should not be specified in k=10 variant

This extra attribute: https://github.com/incf-nidash/nidmresults-examples/blob/master/ground_truth/cluster_k%3D10/nidm.ttl#L15 should be removed from the ground truth as the actual value will vary depending on the dataset.

Ground truth to be updated

This issue os open to keep a track of updates that need to be implemented for the ground truths:

• F-test and T-test should also include a 'Statistic Map' with the corresponding 'statistic Type'
• 2 t-tests ground truth
• in the thresholding ground truths, the threshold should be defined as the main threshold (i.e. used by an inference activity)
• 'spm_covariate', 'spm_2_t_tests' and 'spm_explicit_mask' are currently ignored in the test of the SPM exporter.

Configuration of FSL FIR example

As discussed with @nicholst, the configuration of the FSL FIR example (https://github.com/incf-nidash/nidmresults-examples/blob/master/fsl_hrf_fir/design.fsf) should be reconsidered.

F-test example should not use explicit mask

Currently the F-test example is using an explicit mask:

• https://github.com/incf-nidash/nidmresults-examples/blob/master/spm_f_test/batch.m#L230

@AlexBowring: could you re-run this example without the explicit mask? Thank you!

Recompute examples with FSL 5.0.11

We should recompute FSL examples with the latest release of FSL (5.0.11).

Run BET for all FSL examples

No "stats" folder in "fsl_gamma_basis"

Hi @AlexBowring,

It looks like the stats folder is missing in fsl_gamma_basis. Could you have a look and upload if you have it locally? Thank you!

Document testing variant spreadsheet

@AlexBowring: can you add a page in the Wiki of this repository (cf. https://github.com/incf-nidash/nidmresults-examples/wiki) with a link to our spreadsheet listing the testing variants?

Group t-test with age covariate in SPM and FSL

Now that we are using a dataset for which the subject age is available, it would be best to use age as a covariate in SPM and FSL (rather than subject number). This is already implemented for AFNI.

Delete unused files

The following files are duplicates and should be deleted from this repository:

• #nidm.ttl# in https://github.com/incf-nidash/nidmresults-examples/tree/master/ground_truth/voxel_FWE_p_05
• VUncorrectedpdf.m in https://github.com/incf-nidash/nidmresults-examples/tree/master/spm_voxelwise_p0001

typo: fsl_full_examples001 should be named fsl_full_example001

FSL examples use 2 contrasts

Currently all of the FSL examples have been exported using 2 contrasts

FSL analyses with registration to standard space

Explicit mask not run

Currently the NIDM export of the "explicit mask" test case cannot be performed because the file is not available in the repo. To solve this issue, we need to:

• Include the nifti file used as an explicit mask in the repo (e.g. named "explicit_mask.nii")
• In nidm_export, re-write path to the explicit_mask.

Version should be linked to ground truth rather than test data

cf. for example https://github.com/incf-nidash/nidmresults-examples/blob/master/fsl_FWE_p005/config.json#L6

AFNI analyses with 6mm smoothing

Ground truth file for all examples

We need to create a config.json file (and a ground truth file if not already existing) for all the new examples (including the AFNI ones).

Add namespace prefixes in nidm.ttl ground truth voxel_FWE_p_05

Namespace prefixes are missing in ground_truth/voxel_FWE_p_05/nidm.ttl. Similarly to what was done in ground_truth/voxel_uncorrected_p_001/nidm.ttl#L1-L4, @prefix should be added at the beginning of the file to define nidm:, prov:, rdfs:...

Missing AFNI ground truth files

Some config.json and ground truth files are missing for the AFNI analyses (e.g. afni_gam)