Both need a separate measurement of the B0 field/distortions. FUGUE requires a gradient echo field map which measures the perturbations in the field caused by the subject. This is used to then correct the data as we have a theoretical mapping between the field perturbations and EPI distortions.

Topup requires non diffusion weighted Spin Echo EPI  (EPI Fmap) which are otherwise matched to your acquisition. Moreover, you need atleast two orthogonal phase encoding directions for your EPI fmaps for Topup to work correctly. With this method you are directly measuring the distortions rather then measuring the field perturbations and then computing distortions.

Either method requires extra data and they both have their pros and cons. With a GRE Fmap, you only need one acquisition which can be shared between DWI. FMRI etc. With topup, the EPI fmaps have to be matched to your acquisitions. However, the topup method provides better distortion correction as you end up directly measuring distortions rather than computing it.

To answer your question. No Fugue and Topup are completely separate tools that perform the same task very differently. You only need to use one or the other.

Germaine,

if i'm not wrong, FUGUE overcomes EPI distortions by using a sequence called field-map (which is normally acquired prior to bold-related sequences). TOPUP instead corrects such distortions by aligning two sets of diffusion-weighted images acquired in opposite directions (anterior-posterior and posterior-anterior, for example). 

If you neither have a field-map sequence nor opposite-encoding directions for your DWI dataset, you can solve this by registering your b0 volume into a T1-weighted image in BrainSuite software.

The later solution (BS) maybe it's not accurate as the former ones, but it sure does better than doing nothing.

cheers,

J

Hey Jonatan!

If i did acquire the fieldmap and DWI, does this mean that I can first use TOPUP, then eddy_correct followed by FUGUE before further pre-processing?

Is there a fixed step that one must follow or i can actually do any of them first?

Thank you!

Germaine

Both need a separate measurement of the B0 field/distortions. FUGUE requires a gradient echo field map which measures the perturbations in the field caused by the subject. This is used to then correct the data as we have a theoretical mapping between the field perturbations and EPI distortions.

Topup requires non diffusion weighted Spin Echo EPI  (EPI Fmap) which are otherwise matched to your acquisition. Moreover, you need atleast two orthogonal phase encoding directions for your EPI fmaps for Topup to work correctly. With this method you are directly measuring the distortions rather then measuring the field perturbations and then computing distortions.

Either method requires extra data and they both have their pros and cons. With a GRE Fmap, you only need one acquisition which can be shared between DWI. FMRI etc. With topup, the EPI fmaps have to be matched to your acquisitions. However, the topup method provides better distortion correction as you end up directly measuring distortions rather than computing it.

To answer your question. No Fugue and Topup are completely separate tools that perform the same task very differently. You only need to use one or the other.

Germaine,

you should choose just one between FUGUE or TOPUP. If your DWI dataset were acquired in two opposite directions, I would definitely go with the topup's option (Andrew Jahn explains it very well in https://www.youtube.com/watch?v=zp0EOuKO3O0). As Parnesh Raniga said, topup slightly outperforms fugue's correction.

below in the image attached there is an example of EPI correction I conducted using a T1-weighted image (this is the worst case scenario approach) in BrainSuite. As you may see, deformations are still visible but the overall image improved a little.

the preprocessing order in FSL should be (1) correct EPI distortions and (2) apply eddy-current and motion correction (and bvec file rotation, if you're running the old eddy_correct instead of the new eddy command). As far as I know, unwarping distortions in step 1 introduce anatomical modifications that can impact bvecs.

In BS is different because it rotates automatically your bvecs file after EPI correction. So in this case, you probably should correct eddy currents and head motion first!

hope this helped you,

J

Thank you Parnesh and Jonatan! :)

Hi Jonatan and Parnesh, 

I am trying to run my eddy corrected data through FSL FUGUE with the following information(see pictures attached). However, the TR(s) changes from 6.8 to 1.0. I was wondering if you would know this change will affect the data? 

Also, what should i put for high pass filter cutoff given that i am doing a diffusion analysis. Will the default (100) affect the diffusion data? 

Thanks for your help!

Hi Jonatan Ottino-González

what function do you use to correct an EPI image with a t! unsing BrainSuite? Thank you for your help

hi Daniel Roquet, I call the bdp function (http://brainsuite.org/processing/diffusion/flags/) as in the following example:

/Applications/BrainSuite16a1/bdp/bdp.sh T1_image_bfc.nii.gz --tensor --nii dwi_image.nii.gz -g dwi_bvecs -b dwi_bvals

If you're running this in Windows SO remember to put the path to the bdp function (/Applications/BrainSuite16a1/) between commas ("/Applications/...").

It is also important that your T1 file (already skull extracted and corrected for intensity inhomogeneities) has the "bfc" acronym somewhere in its name, otherwise, BS will fail to recognize your T1-image.

This function (~5 minutes per subject) will generate a bunch of files, but you're only interested in the one ending in ".dwi.RAS.correct.nii.gz". That's your DWI volume corrected for EPI distortions.

If by any chance you want to use the other diffusion maps for your analysis you also can (e.g., FA, MD, AD, RD). Only have in mind that they're in T1-resolution (usually 1x1x1), which is not the one they were initially in (i.e., 2x2x2).

Hope this has helped you.

Best,

J