I am new to MRtrix and attempting to adapt a pipeline written by a colleague from another lab. When I ran dwi2fod on a dataset I got a set of errors saying “dwi2fod: [INFO] voxel [37 69 52] did not reach full convergence” although there was a set of errors, each with different voxels locations. I couldn’t see any obvious distortion or signal drop out in the FOD image.
I can’t find any documentation relating to this error. Is it serious? Can I correct it?
Hi @RachelSharkey,
Was this using the csd or msmt_csd algorithm? Using the older ‘csd’ algorithm, these errors are not all that uncommon, and usually nothing to worry about much – it typically means that the solver is oscillating between two solutions, both of which are usually pretty close anyway.
If using msmt_csd, it’s not something I’ve seen happen very often, though it is also possible. The best thing to do is to look at the affected voxel in mrview and check whether the estimated FOD is reasonable. You may actually find it’s not in a location you care about anyway (I typically see these in non-brain voxels that happen to be included in the mask).
All the best,
Donald