Does BIDS currently have a way to indicate when a scanner applies intensity non-uniformity correction during reconstruction (e.g., GE PURE, Philips CLEAR, Siemens Prescan Normalize)? I couldn’t find a field for this. It seems useful to record it since it can affect downstream preprocessing. Maybe something like:
I believe the term NonlinearGradientCorrection has been proposed but not adopted by BIDS. If you used dcm2niix and a Siemens scanner, you can typically infer this from the ImageTypeText or ImageType fields.
“intensity non-uniformity correction” addresses the fact that the receive coils don’t have a uniform sensitivity (so typically regions closer to the coils are brighter than farther away). You correct for that by normalizing each voxel intensity by the coil sensitivity map.
“non-linear gradient correction” refers to the fact that the magnetic field created by the gradient coils is not perfectly linear with distance from isocenter, while the image reconstruction assumes it is. So you end up with minor distortions in the image (there will be more distortions the further the deviations from the ideal linearity of the field). If you know the exact magnetic field mapping for each of the X, Y and Z coils, you can correct for those distortions (more or less*).
*More or less because the phase of the image might have been discarded before applying the non-linear gradient correction, but the effect should be small.