How to deal with "stripe" artifact in fMRI?


I was wondering if anyone knows how to deal with this kind of stripe artifact accompanied with ghosting in fMRI (see attached figure) ? The stripe pattern persists after ICA-AROMA denoising. Do I have to abandon this data or is there any way to rescue it?

Thank you very much!



I do not think that what you see is due to ghosting. I would think it is linked to cross talk between slices. What kind of acquisition (Magnetic Field, in-plane resolution, slice thickness, slice gap, TR, slice acquisition order, multi-slice, continuous or sparse sampling) is it?

It might be interference cross-talk or head-motion related spin history artefacts. The former would be similar in magnitude in every volume, while the latter would be more salient in some volumes than others. Interleaved acquisitions are more resilient to the former, while sequential acquisitions are more resilient to the latter. Adding a bit of a slice gap is the solution to both problems. See the Common Artefacts page.

Hi @neurolabusc @jsein
Thank you very much for the answer. I was wondering if there is some post-processing strategy to rescue the slice cross-talk problem or the solution must be by modifying acquisition parameters at the scanner?

MRI processing usually includes a Gaussian blur with a full-width half maximum of around 2-3 voxels. This should minimize some of these artifacts. I suspect clever solutions from the 2D domain could be adapted to the 3D domain.

1 Like