Closed AndrewLawrence closed 6 years ago
I confirm this is a bug. It will need some investigation - the residual output is allocated in the same way as the parameter outputs, so it is not immediately clear why this is happening.
Hello, This should be fixed in the development branch (commit d9bb191). The DESPOT residuals were being divided by the Proton Density image before writing, so they were of comparable magnitude. This was causing a 0/0 = Inf error outside the mask. I've swapped to simply writing out the unscaled residuals.
Residual images from qi tools (qidespot1hifi, qidespot1, qidespot2fm, qimcdespot) have a uniform single large value outside the provided mask. It is probably max float: ~3.4×10^38.
To view the residuals I have first been masking in FSL:
fslmaths HIFI_residual.nii -mas Mask.nii HIFI_residual
Although this step is pretty simple, it would be more user friendly to output zeros or NaNs?