Simulation of common MRI artifacts

,
1

Hi everyone,

I’m looking for tools/libraries which enable the simulation of artifacts in MRI acquisitions. Anyone familiar with something of the sort? Maybe @effigies @oesteban @satra @neurolabusc ?

Any relevant resources would be greatly appreciated.

Thank you very much in any case :pray:

1 Like
2

Hi @ZviBaratz,

Maybe FSL POSSUM? See Drobjnack et al 2006, 2010, and Graham et al., 2016.

Best,
Steven

1 Like
3

indeed possum can help. there is some work that both francisco’s group at nimh and doug’s group at mgh were doing in the last few years on this topic. i don’t think either are on neurostars. also, i saw a nice presentation of a simulator from a group in netherlands during ismrm last year. you may want to check the abstracts.

1 Like
4

I agree POSSUM seems like the best option at first.

For dMRI, you may want to have a look into Fiberfox https://doi.org/10.1002/mrm.25045

1 Like
5

not artefacts, but BrainWeb helps generate MRI scans which might be useful to test on noise levels IIRC

1 Like
6

Thank you! These are great leads.
I have to admit that I find POSSUM’s legacy installation status somewhat worrying, but it certainly seems like a strong choice.
Also, @raamana, thanks for also mentioning BrainWeb. I wasn’t familiar, and it is a very interesting project.

7

you’re welcome. i had been wanting to spend more time in that direction but was unable to, so I am happy to be learning more about and contributing to what you and others in this space would achieve.

an interesting addition to VisualQC would be to improve visualizations to enable the detection of subtle artefacts in a faster and more reliable way: for example, we saturate an anatomical image to reveal any noise or artefacts in the background:
https://raamana.github.io/visualqc/gallery_t1_mri.html#gallery-structural-t1w-mri-artefact-detection-and-rating

so if we know how to generate artefacts, we can also use that fact to improve how we detect them :slight_smile:

1 Like
8

A deeper understanding of k-space can often help one understand artifacts. Many simulators exist, and ‘extreme’ values of various parameters sometimes give insights: https://kspace-explorer.streamlit.app/

2 Likes
9

This app is wonderful! Thank you.

1 Like
10

@ZviBaratz If you want the most realistic MRI simulator, I can highly recommend checking out Corsmed: https://www.corsmed.com/

Corsmed creates every image from scratch, so whenever you make the slightest change in any input setting, that change is captured exactly in the resulting image.

Corsmed is like having a real MRI scanner on your laptop – but one that is available 24/7 whenever you need it, and at 1% the cost of using a real scanner.

On Corsmed, you can also:

  • Choose any part of the body to scan – brain, cardiac, lumbar spine, breast, liver, chest, etc
  • Select what coils and magnetic field strength to use on the scanner
  • Choose among 25 different pulse sequences
  • Plan how the slices are angled
  • And for every sequence, you can specify more than 25 different parameters, including Field-of-View (FOV), Matrix, Number of Slices, Slice Gap and Thickness, Echo Time (TE), Repetition Time (TR), etc.

Whatever scan you want to simulate, you’ll be able to do it on Corsmed – and get just as accurate images as on a MRI real scanner.

I hope this helps! :slight_smile: