Print

Print
Thank you so much for your help and the suggestions. I will have a look at the acquisition parameters. Also, performing multiple BET's sounds like a good solution. 

I only really use the anatomical scans for registration purposes. I was under the impression that in order to get good registration it is important to virtually eliminate all non-brain tissue in structural brain extraction? Or is this being overly meticulous?

Laura

On Aug 6, 2015, at 1:53 PM, Mark Jenkinson <[log in to unmask]> wrote:

Hi Laura,

With this acquisition it is tough to get everything that you'd want.
I have a fairly good result with:
  bet rfov.nii.gz rfov_brain -R -f 0.3 -g -0.2
on the result from robustfov (I've called it rfov here).
However, it still includes non-brain material at the top, which is largely because you have quite a lot of signal from the marrow/membranes that is right next to the brain and it is difficult for BET to separate them cleanly.  BET tends to work better when fat saturation is applied (at least partially) in the structural image acquisition. The frontal lobe is difficult because you have a fairly local bias field there that is making it quite dark.  The results of the above look pretty good for me in the frontal lobe, but I had to use -g to stop it running away in the inferior regions. See what you think.  For the purposes of registration this should be sufficient.

If you need a very tight brain mask then I'd suggest running two BETs, one optimised to get the bottom half and frontal lobe right, and then one optimised to get the top portions right, and then just combine the two together with fslroi and fslmerge.  This may even work as a general solution if you have lots of images like this and script up the necessary commands.  I can't think of anything else to suggest at the moment.

All the best,
Mark

From: FSL - FMRIB's Software Library <[log in to unmask]> on behalf of Laura Banu <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Thursday, 6 August 2015 18:31
To: "[log in to unmask]" <[log in to unmask]>
Subject: Re: [FSL] Problematic Brain Extraction

Thank you so much for your help, as always. I will upload the original image now (245016_Structural)

---
Laura Banu
Research Assistant
Cardiac Neuroscience Laboratory
Center of Alcohol Studies
Rutgers University
Work: (848) 445-3577
Cell: (646) 643-6917

On Thu, Aug 6, 2015 at 1:25 PM, Mark Jenkinson <[log in to unmask]> wrote:
Hi,

If the output of robustfov does not look obvious wrong then I doubt -b will do anything useful.  Can you upload your original image to:


I'll then have a look and see if I can see why there is a problem and how to fix it.

As for your other question, the answer is yes.  You would use the reduced FOV output from robustfov in all subsequent analyses; as if the original one never existed, as we never need the neck, unless you are doing spinal cord imaging, where it is quite helpful!  ;)

All the best,
Mark


From: FSL - FMRIB's Software Library <[log in to unmask]> on behalf of Laura Banu <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Thursday, 6 August 2015 18:21

To: "[log in to unmask]" <[log in to unmask]>
Subject: Re: [FSL] Problematic Brain Extraction

Unfortunatly, I am still not getting a good output. I will try the -b option next.

Another question: if I use robustfov for brain extraction, and then use BBR for registration later on, I'm assuming the non-brain extracted input for BBR should be the output from robustfov?

---
Laura Banu
Research Assistant
Cardiac Neuroscience Laboratory
Center of Alcohol Studies
Rutgers University

On Thu, Aug 6, 2015 at 1:09 PM, Mark Jenkinson <[log in to unmask]> wrote:
Hi,

You don't need the -f and -c options.
The rest looks good.

All the best,
Mark

From: FSL - FMRIB's Software Library <[log in to unmask]> on behalf of Laura Banu <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Thursday, 6 August 2015 18:04
To: "[log in to unmask]" <[log in to unmask]>
Subject: Re: [FSL] Problematic Brain Extraction

Dear Mark,

Thank you for the speedy reply!

I will give this a try. Just to make sure I understand, the commands would look something like this?

robustfov -r 245016_RFOV -i 245016_Structural
bet 245016_RFOV 245016_Structural_brain -f .45 -c 78 130 110 -R

or, do I no longer need to specify the -f and -c options?

Laura

---
Laura Banu
Research Assistant
Cardiac Neuroscience Laboratory
Center of Alcohol Studies
Rutgers University

On Thu, Aug 6, 2015 at 12:20 PM, Mark Jenkinson <[log in to unmask]> wrote:
Hi,

We have found that the following prescription works quite well:

 - run robustfov to save out an ROI (-r option) without any neck
 - run bet with the -R option

Depending on your subjects and alignment the first stage may be too tight in the z-direction, in which case you should alter the size of brain in z (the -b option) but in general it works well.

Give it a try and let us know how you get on.

All the best,
Mark


From: FSL - FMRIB's Software Library <[log in to unmask]> on behalf of Laura Banu <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Thursday, 6 August 2015 16:24
To: "[log in to unmask]" <[log in to unmask]>
Subject: [FSL] Problematic Brain Extraction

Dear experts,

I have been working on a particular brain extraction for quite some time and can't seem to get a good result. The most recent parameters I tried are as follows:

bet 245016_Structural 245016_Structural_brain -f 0.45 -c 78 130 110 -B

No matter what I try, there seems to be a fair amount of frontal lobe missing and a fair amount of non-brain at the top. Any advice is greatly appreciated.

Thanks,
Laura

ps. Sorry if this message posted twice. I tried to attach the files, but they were too large