Neonatal skull stripping

Hi,

I didn’t have any problems with that. Maybe I had already the 2 versions of cmake and I don’t know.

Best regards,

Manuel

So do you think using the default templates included with MASS would produce bad results?
thanks,
Rosella
PS: I successfully installed ANTs but I still need to install AFNI in order to correctly use MASS.
Best,
Rosella

Hi,

The default templates are adult subjects. To be fair, I don’t remember even trying it, I directly created my own neonatal templates, I expect to result in much accurate masks.

Best regards,

Manuel

Sorry,
in particular, which data from dHCP project did you use as template?
So this way I am sure I will get good results.
Many thanks,
Rosella
PS: I managed to install everything properly finally!

Hi,

I don’t use the dHCP data, I have my own cohort processed with the dHCP pipeline.

You should select 6 subjects (for example) and create a folder where you rename the files:

*_T1w_restore.nii.gz -> Template*.nii.gz
*_brainmask_drawem.nii.gz -> Template*_str_cbq.nii.gz

I would check that the brains and the masks overlay correctly in mrview and fsleyes.

Best regards,

Manuel

1 Like

So do you suggest to select 6 neonatal subjects from dHCP data release and use them as a template _ or it is better to use my own subjects?
thanks
rosella

Hi,

If you have accurate masks for som of your subjects, I would say that is better to use that ones.

Best regards,

Manuel

What I can’t understand is that I do have neonatal subject data but these the images I need to skllstrip. So should I create these mask manually ?
best,
Rosella

Hi,

Yes, the subjects of the template folder should have a mask associated. If you don’t have it, you can use the subjects of the dHCP.

Best regards,

Manuel

1 Like

In case, should I create these associated masks manually if I wanted to use my subjects?
thanks,
Rosella

Hi,

You can create that masks as you like. For example with bet or robex and manually edite it. Take into account that the quality of your mass generated masks depends of the mask you have created for the templates.

Best regards,

Manuel

So basically you took your own 6 subjects and processed them with dHCP pipeline and used the output as reference template for mASS?
thanks,
Rosella

Hi. I have got a problem of compatibility between gcc version 5 required by dHCP pipeline installation and my OS system (Debian Jessie). Would you please be so kind to give me your own templates you used or If I give you 6 of my subjects create these templates for me ? I would include your name in my paper.
Thank you anyway for your availability,
thanks,
Rosella

Sorry,
I performed FOV reduction, bet and bias field correction on ALL my 3dT1 images and I am currently running mass using the 6 templates you gave me as reference. I can’t understand however the last 2 commands you provided me here .
Thanks,
Rosella

Hi.
I have been checking skull stripping on all of my subjects and I can see issues in some of them, the rest instead gets pretty results. Do you have any hypothesis on the reason why?
thanks,
Rosella

Hi,

For the first image, I would say that is because there are articatc in the frontal area and the registration failed, and for the second one, it looks like the FOV reduction was excesive. Could you send a couple of screenshots of the original images for this subjects?

Best regards,

Manuel



thank you

Hi,

For the second subject, I would say tory to repeat it removing the robustFOV step. For the first one, I don’t know what could be going on. Try to inspect the outputs of each step.

Best regards,

Manuel

1 Like

I think the problem with the second subject is at bias field correction step. The output of bfc, after a first run of FOV reduction and bet, is:

Hi. I haven’t solved the skull stripping problem with those 4 subjects where N4 bias field correction seems to fail. I have no idea of how solving this problematic step . Could skipping bias field correction be an option? and directly applying mass? I was wandering if it is correct for subsequent analysis to just do this on 4 subjects.
Best,
Rosella