We tend to use sigloss for a fairly rough calculation of signal loss (since it doesn't take into account within-plane sources of signal loss) but to make it "more accurate" the fieldmap that it is run on should have voxel sizes which match the EPI (a resampled fieldmap is fine for this). Smoothing the fieldmap in the slice direction will have a fairly big impact on the signal loss so I would avoid that if you want more "accurate" answers. However, for getting a rough indication of which parts of the brain to ignore in registration, via a cost-function weighting, which is what we use sigloss for, then it doesn't matter so much.
All the best,
On 29 Mar 2013, at 10:50, Basile <[log in to unmask]> wrote:
> Hi fsl users,
> I do have questions about sigloss:
> - Does it have to be run on the resolution of the image in which we want to estimate the signal loss or the original fieldmap resolution or any?
> I mean the signal loss is computed using the gradient of fieldmap in slicing axis (default z) so if the fieldmap is resampled to another space (let say T1 for instance), sigloss values will be different than in it's original space.
> - If the fieldmap has been smoothed in the slicing direction, will it cause problems in the estimate of signal loss too?