Gibbs ringing removal

Dear experts

I was just wondering:

  • whether Gibbs ringing removal (Kellner et al., 2016) has been implemented in mrtrix?

  • whether intensity normalization is required for ACT using multishell data?

Thanks and cheers,

Not yet, but it’s almost there - just need someone to push a bit harder, and we’ll get it done… I think there’s enough demand for it, both internally and externally, to try to prioritise this - I’ll see if I can find the time to polish off what @bjeurissen’s has already done.

No, not as far as I know (@rsmith might like to comment here). But you’d well advised to run at least bias field correction. If you intent applying SIFT / SIFT2 on the tractograms and comparing the results across subjects, then intensity normalisation will be important for accurate quantification.

If you’re only performing ACT, which is by definition a single-subject framework, then inter-subject intensity normalisation is not going to have any effect (over and above the scaling of FOD amplitudes, which in turn may influence streamlines terminating prematurely due to the FOD amplitude threshold). If you wish to compare connection densities between subjects (which is made more robust through use of ACT, but isn’t a constituent part of ACT), then one of the ways that you can choose to normalise these connection densities across subjects does involve the use of AFD-like intensity normalisation.

Hi @Hamed,

It’s probably worthwhile mentioning this is now available publicly since last week’s update. If you update your installation, you should be able to make use of the new mrdegibbs command. I’ve just tested it over here on some of my data, and it works like a charm!


Thanks @ThijsDhollander and team,

Can you recommend at any particular stage to remove the Gibbs artifacts?

I figure it would be one of the initial steps you do, but before the denoising (i.e. dwidenoise)



Hi Alistair,

Always do the denoising first. We recommend doing Gibbs ringing removal directly after denoising, before motion correction or any other image interpolation.


1 Like

Hi This

Thanks for the update.
I went on with unring.a64! But will definitely give it a shot for another study. :slight_smile:


1 Like