Christopher J Markiewicz

Yeah, ideally it would be to put a limit on the maximum distance of the generated warp field (I'm assuming that ANTs stores warps as arrays of vector displacements). Even better would be to add a per-voxel cost, such that a voxel with a lower displacement prior is penalized more heavily for a greater displacement, while a voxel with a higher prior would have less penalty, because we expect it to require greater distortion. I'm sorry if I'm being unclear; I'm not well-versed in non-linear registration...

Hi, I'm working on estimating the warp of an EPI image to a skull-stripped T2 (or inverted T1) in the phase-encoding direction, as a way of approximating fieldmap unwarping in situations where the data is unavailable (following on similar work by Huntenburg and Wang et al.). One thing I'm looking into is using an average fieldmap warp atlas as a prior. As a first pass, I'm using this to generate a mask to limit the SyN step, but is there a way to also use it to constrain the distance of displacement...