Any way to work around a chemical shift artefact?


Hello MRtrixers,

I have noticed that most of my diffusion scans unfortunately have a chemical shift artefact. Here’s an example of such artifact:

The artefact is also quite obvious in the ODF map:

Once I generated streamlines for the whole brain, the artefact wasn’t as obvious, but I’m sure it still affected the tracking process. Here’s what it looks like:

My question is, is there any way I can work around this artefact using MRtrix? I’m trying to track the optic radiations, which will obviously be affected by the artefact, but how badly will it do so? I know my question is quite vague, but I’m honestly just looking for suggestions on how to handle this data. Is the data even salvageable?

As always, thanks for the help.



I’m not sure there’s a simple way to remove this kind of artefact. Part of the problem is that it’s typically incoherent, it can add or subtract from the signal randomly due to the extreme sensitivity to motion of the DW signal, and the fat signal can be relatively large compared to the DW signal at high b-values. I really don’t know of any method do deal with this… I’m not sure this necessarily means you need to throw out your data though – fat artefacts are fairly common in DWI. I reckon it just means you need to be careful with any findings that involve this region…