search

vsoch / nidmviewer

NIDM Results Viewer
http://vsoch.github.io/nidmviewer/
BSD 3-Clause "New" or "Revised" License
3 stars 3 forks source link

Display statistic map even if excursion set is empty #10

Closed cmaumet closed 8 years ago

cmaumet commented 8 years ago

This PR fixes #9 by enabling display of a statistic map in the absence of suprathreshold voxels.

Tested on my machine the following command:

nidmviewer PATH_TO_EXAMPLE/fsl_ds005_group/nidm.ttl

returns: image

cmaumet commented 8 years ago

(Note as FSL does not currently return the list of peaks and clusters for voxel-wise thresholds, the list of peaks can be empty even if the excursion set is not).

chrisgorgo commented 8 years ago

Since we are moving to displaying excursion sets, wouldn't a better approach be just to display a warning saying that "there are no suprathreshold voxels" and not display any map? It would be more consistent.

vsoch commented 8 years ago

What exactly is an excursion set? If it is what I think it is, +1

chrisgorgo commented 8 years ago

Excursion set is a fancy name for a thresholded map (as in a nifti file with non-significant voxels set to to zero).

cmaumet commented 8 years ago

@chrisfilo: as FSL does not report peaks and clusters for voxel-wise threshold, you might have no peaks and clusters even if the excursion set in non-empty.

chrisgorgo commented 8 years ago

That sounds suboptimal and limits the usefulness of NIDM-Results for people doing peak based metaanalysis.

In the context of the current bug: Is there a way in NIDM-Results to reliably (across FSL, SPM, AFNI) check if there are any suprathreshold voxels without having to read any of the nifri files?

cmaumet commented 8 years ago

That sounds suboptimal and limits the usefulness of NIDM-Results for people doing peak based metaanalysis.

Yes, I agree. We are working with FSL to make this happen. But this is not a limitation of NIDM-Results as any study using a voxel-wise threshold within FSL will have no listing of peaks and clusters.

In the context of the current bug: Is there a way in NIDM-Results to reliably (across FSL, SPM, AFNI) check if there are any suprathreshold voxels without having to read any of the nifri files?

I don't think there is a way. Shall we continue with the current approach and leave the update to display "no significant clusters" when appropriate for when the issue with FSL is sorted?

chrisgorgo commented 8 years ago

Saying "no significant clusters" when the excursion set is not empty is very confusing. I would prefer to avoid this.

@vsoch Can the viewer check if the excursion file includes any nonzero non-nan voxels?

cmaumet commented 8 years ago

@chrisfilo: Yes! That's why if we can't know for sure I think we should just not display any message.

chrisgorgo commented 8 years ago

Not displaying any message when the table and the excursion set are both empty is also confusing. It looks like the viewer is broken.