Hi Lili,
There is no need to replace any files. You can just:
1. Use fsl_reg <your T1> <your template> T1_to_template -a (You will probably want to run bet on your template and input image first)
2. Use fast with -a T1_to_template -A <your priors>
Hope that makes sense!
Cheers,
Eelke
> On 26 Aug 2015, at 14:37, Lili He <[log in to unmask]> wrote:
>
> Hi Eelke,
> Thanks for the explanation. However, I am trying to segment neonatal brains, so that I cannot use default adult template. I have neonatal brain template and tissue probability maps. I think I can either replace the default adult templates and priors or I can specify which ones I want to use in the command line. It seems FSL only allow to specify which priors to use but not allow to specify which template to use.
> Thank you,
> Lili
>
> On Wed, Aug 26, 2015 at 9:07 AM, Eelke Visser <[log in to unmask]> wrote:
> Hi Lili,
>
> There is no need to specify the actual template - you only need to specify it when running flirt. The default template used by pretty much all the FSL tools is $FSLDIR/data/standard/MNI152_T1_2mm.
>
> Just to be sure. You will need to run these two steps:
> 1. Registration: If you want to use the default priors, the easiest way is to run fsl_reg with its -a option and $FSLDIR/data/standard/MNI152_T1_2mm as the reference image. If you want to use your own template and priors, you can use that template as the reference image instead.
> 2. Segmentation: In this step you just need to specify the matrix you obtained in step 1 with fast’s -a option and optionally your own priors using -A.
>
> Cheers,
> Eelke
>
>
>
> > On 26 Aug 2015, at 13:52, Lili He <[log in to unmask]> wrote:
> >
> > Hello Eelke,
> > I found that option -a only takes <standard2input.mat> as input, while I cannot specify which standard template I want to use. Is there any solution on this?
> > I think you said that FSL uses MNI152lin_T1_2mm_brain as the default template. Is it located under $FSLDIR/data/standard/?
> > Thank you,
> > Lili
> >
> > On Tue, Aug 25, 2015 at 11:05 AM, Eelke Visser <[log in to unmask]> wrote:
> > Hi,
> >
> > Yes, the affine registration matrix that you supply is used to register the images. Normally you would get this matrix by manually using flirt to register your data to the MNI template, i.e. MNI152_T1_2mm_brain. An easy way to do this is using the fsl_reg tool. The avg152T1_brain image is a copy of MNI152lin_T1_2mm_brain, which is the same template created using linear regstration as opposed to nonlinear. You can use either one as your flirt target - it shouldn’t make much difference.
> >
> > If you want to use your own template and priors, there is no need to replace any files in the FSL directory. You can just use flirt to register to your own template instead and provide that matrix using the -a option. You should then specify your own prior images using the -A option. If your images are already registered, you can disable registration by specifying $FSLDIR/etc/flirtsch/ident.mat as the transformation, though it is probably best to run FAST on the original images and provide the matrix.
> >
> > If you are using multichannel FAST, you can use the T1 to register to your template (assuming that is generated from a T1) and register each subject’s T2 to that subject’s T1 image.
> >
> > Cheers,
> > Eelke
> >
> >
> >
> >
> > > On 25 Aug 2015, at 15:17, Lili He <[log in to unmask]> wrote:
> > >
> > > Thank you Eelke! It is very helpful!
> > > Under $FSLDIR/data/standard/tissuepriors, I saw 4 files. My understanding is FSL first register my input image to the avg152T1_brain.img (while I am not very sure about this, or may the other atlas located under fsl/data/standard?), and then use avg152T1_white.img, avg152T1_csf.img, avg152T1_gray.img as the a prior maps for initial parameters or/and for final segmentation. The registration is conducted according to the affine transformation matrix I will provide. Is this correct?
> > >
> > > If this is what FSL does, I believe that I can put my own a prior maps (neonatal ones in my case ) under the folder of $FSLDIR/data/standard/tissuepriors. And I can also put neonatal structural template in this folder or where should I put my neonatal structural template?
> > >
> > > I understand if my input image is not normalized into the template space, I have to provide the transformation matrix to have fsl do the registration. However, if I have my input image already registered into the template space, how will I let fsl know not to do the registration? Simply leave the transformation matrix empty?
> > >
> > > In addition, if I use multi-channel segmentation, say using both T1 and T2 images. Which template and which input image (T1 or T20) will be used to for the registration? since I only saw a T1 template under $FSLDIR/data/standard/tissuepriors. Or I guest what I want to know is how I can specify which template to use, since I saw different atlas under fsl/data folder.
> > >
> > > Please let me know if my expressions of the questions are not clear, I will try to clarify. Thank you so much for your help!
> > > Lili
> > >
> > >
> > >
> > > On Mon, Aug 24, 2015 at 5:43 PM, Eelke Visser <[log in to unmask]> wrote:
> > > Hi Lili,
> > >
> > > You can use the -a option to enable the use of priors (it requires you to specify the affine registration matrix that registers your data to MNI space). This will use the priors in $FSLDIR/data/standard/tissuepriors - you can specify your own using the -A option in addition.
> > >
> > > The default is to use the priors only for initialisation (to get an initial guess for the mean intensity for each tissue type). You can use the -P option to tell FAST to use them during the actual segmentation stage as well. This is what the manual says (this refers to the GUI):
> > >
> > > "Use a-priori probability maps tells FAST to start by registering the input image to standard space and then use standard tissue-type probability maps (from the MNI152 dataset) instead of the initial K-means segmentation, in order to estimate the initial parameters of the classes. This can help in cases where there is very bad bias field. By default the a-priori probability maps are only used to initialise the segmentation; however, you can also optionally tell FAST to use these priors in the final segmentation - this can help, for example, with the segmentation of deep grey structures."
> > > (see http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FAST)
> > >
> > > Hope that helps!
> > >
> > > Cheers,
> > > Eelke
> > >
> > >
> > > > On 24 Aug 2015, at 19:46, Lili He <[log in to unmask]> wrote:
> > > >
> > > > Dear FSL experts,
> > > > I was trying to run multi-channel FAST for tissue segmentation. My question is how I can input a prior probability tissue maps (something like grey.nii, white.nii and csf.nii in SPM ).
> > > > Thank you,
> > > > Lili
> > > >
> > >
> >
> >
>
>
|
