Noise FWHM in nidm namespace

[cmaumet]

This is a proposal to move the "noiseFWHM" terms (i.e. spm:'noise FWHM in units', spm:'noise FWHM in voxels' and spm:'noise FWHM in vertices') to the nidm namespace.

We initially created those terms in the SPM namespace because those values were only reported for SPM results. But:

• Since fsl, spm and nidm namespaces #301, we have considered that terms representing generic concepts would be defined in the nidm namespace.
• Moreover, issue Harmonise FSL and SPM noise FWHM terms #319 and Refining the definition of FWHM, and defining FSL's equivalent measure #214 (cf. in particular Refining the definition of FWHM, and defining FSL's equivalent measure #214 (comment)) have proposed to also report 'noise FWHM in units' and 'noise FWHM in voxels' for FSL (implemented in cmaumet/nidmresults-fsl@82c51f9).

[nicholst]

+1

[khelm]

Do the terms have the same definition in SPM and FSL?

[cmaumet]

Yes, current definitions (for SPM) are generic and could also be used for FSL:

• spm:'noise FWHM In Voxels': "Estimated Full Width at Half Maximum of the spatial smoothness of the noise process in voxels"
• spm:'noise FWHM In Units': "Estimated Full Width at Half Maximum of the spatial smoothness of the noise process in world units"

[khelm]

OK, thanks. +1

[nicholst]

For the record, FSL and SPM use different methods to estimate the FWHM (FSL uses correlation of residuals at pairs of adjacent voxels; SPM uses variance of the difference of residuals at pairs of adjacent voxels), but they are both trying to estimate the same intrinsic thing.

[khelm]

Ok, so in a sense they have different "definitions". This seems to be a case where we should decide an overall strategy for this issue. My view is that this is the kind of info that one would want to capture in the provenance, especially since we (i.e., you) actually know the difference. So in the case where we could write a generic definition to cover all cases, but the underlying method is different, could we use the generic definition language, but then have a second sentence saying what method is used?

[cmaumet]

We discussed this on NIDASH call on May 11th (Minutes).

[cmaumet]

(Rebased on master). I am going to merge this as agreed on NIDASH call on May 11th.