Hi,
These commands all look correct to me.
However, I hope you realise that the FIRST segmentation of the
cerebellum will not separate the gray and white matter *within*
the cerebellum. It simply tries to extract the whole cerebellum
as a single structure.
I believe that 40 modes is reasonable for this and using
no boundary correction is a conservative option. You might
want to also explore the other options on your images and
see what the performance is like, as it will vary with SNR,
contrast and overall quality of your images.
I hope this helps.
All the best,
Mark
On 13 Aug 2009, at 09:57, Takuya OGURI wrote:
> Dear all
>
> I would like to segment gray/white matter of the cerebellum by using
> FIRST.
> Following the FSL webpage, procedures are, I suppose, as follows;
> ('Subject1' is a T1 image.)
>
> first_flirt Subject1 Subject1_to_std_sub -cort
> run_first -i Subject1 -t Subject1_to_std_sub_cort.mat -n 40 -o
> Subject1_R_Cereb -m
> ${FSLDIR}/data/first/models_336_bin/intref_puta/R_Cereb.bmv -intref
> ${FSLDIR}/data/first/models_336_bin/05mm/R_Puta_05mm.bmv
> first_boundary_corr -s Subject1_R_Cereb -i Subject1 -b none -o
> Subject1_R_Cereb_corrected
>
> Are those procedures appropriate for segmentation of the cerebellum?
> Specifically,
> 1. What are suitable number of modes in /run_first/ (Is 40
> suitable)?
> 2. Which is adequate option for -b in /first_boundary_corr/ (fast,
> thresh, none)?
> I would appreciate it very much if anyone could give me comments on
> this issue.
>
> Best wishes,
>
>
> Takuya Oguri
> Department of Neurology
> Nagoya City University
> Japan
>
|