Print

Print
Hi,

it would seem a bit strange to me if RF bias field only affects 10 out of 500 cases.
Were these scanned with a different coil?
Cheers,
Andreas

Von: FSL - FMRIB's Software Library <[log in to unmask]> on behalf of Mark Jenkinson <[log in to unmask]>
Antworten an: FSL - FMRIB's Software Library <[log in to unmask]>
Datum: Montag, 14. September 2015 09:27
An: <[log in to unmask]>
Betreff: Re: [FSL] Artefacts in fMRI data

Hi,

It probably won't make a lot of difference in cases where you don't have much artefact, but you need to test that in order to know for sure.  If you try it on some cases of minimal artefact and see if there is much difference then that will let you decide whether it is worthwhile applying it to all 500 or just the ones where it will make a difference (e.g. 10 or so worst cases).

All the best,
Mark


From: FSL - FMRIB's Software Library <[log in to unmask]> on behalf of KUN HSIEN CHOU <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Saturday, 12 September 2015 11:04
To: "[log in to unmask]" <[log in to unmask]>
Subject: Re: [FSL] Artefacts in fMRI data

Hi Mark

If some of our dataset have this problem [around 10 cases / 500 (whole dataset)], do we need to perform the same correction approach for the rest of the dataset or just use this approach to affected data ?

Best

Paul

2015-09-11 16:19 GMT+08:00 Mark Jenkinson <[log in to unmask]>:
Hi,

The example you show is definitely RF bias field (or B1 inhomogeneity) and not related to B0 inhomogeneity or distortion.  If the temporal SNR in the darker regions is still good then this type of artefact doesn't have a strong impact on the voxelwise analysis.  However, it can affect the motion correction and the registration.  To remove it (assuming you do not have any RF, or B1, mapping scans) then you can calculate the bias field from a T1-weighted structural image using FAST and then apply this bias field (if you output the bias field from FAST then just multiply this with your images) *after* transforming the bias field into the correct space.  If your fMRI and structural images were acquired in the same session then using FLIRT with the -applyxfm and -usesqform (but not -init) options will allow you to transform from one space (the structural space, where you've calculated the bias field) to the functional space.  You can tell if this works by just looking at the output after the transformation and multiplication and seeing if the image looks more uniform.  Do this first, prior to any motion correction, and then it will hopefully eliminate all problems associated with the bias field.

As for the little shop of horrors - this isn't online anymore, as it is out of date.  You are better off looking at the papers about FIX: http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FIX
as these papers show a selection of common artefacts in modern data.

All the best,
Mark

From: FSL - FMRIB's Software Library <[log in to unmask]> on behalf of Sam Rogers <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Friday, 4 September 2015 11:33
To: "[log in to unmask]" <[log in to unmask]>
Subject: [FSL] Artefacts in fMRI data

Hello.

The MELODIC FAQ ( http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/MELODIC/FAQ ) mentions a fMRI little shop of horror, but the link is broken. I have found other resources, but I am not totally sure on this.

I have got some 2x2x2 functional data collected on a 3T scanner where some lateral and posterior areas are much brighter. I guess this is bias field. I've put an example on the web http://postimg.org/gallery/g7q24w9q/

I have field map data, so I can correct for distortion in B0, but I think this is RF related right? What would you suggest best done to resolve this? Would you be concerned with the image quality?

Thanks for helping.
Sam