Hi,
Hopefully this should answer all your questions.
Q1:
The difference between using FIRST with and without bias-field correction
has not been thoroughly tested. From personal experience if the initial
registration works,
the bias-field correction is probably not needed. Having said that, if the
bias-field correction
does a good job then it can't hurt.
Q2:
Yes, FIRST to may be used to create ROIs.
As long as the images from the different modalities are in the same space as
the T1 then the FIRST-generated masks can by applied
to the different modalities.
Q3:
No partial volume effect is considered at this time. Ideally, partial
volume effects should only appear
in the boundary voxels.
Q4:
The output labels should correspond to those on
http://www.fmrib.ox.ac.uk/fsl/first/cma_subcortical_label.html with a
couple
of exceptions. The output of FIRST for a single structure will consist of
2 labels: 1) the label corresponding to "cma_subcortical_label.html" which
represents the interior voxels and 2) the same label as the first plus
100 (i.e. for left hippocampus there exists 17 and 117).
After boundary correction there will exists only the single label.
The output of run_first_all is a 4D file. Each time point after the first
corresponds to each structure with an uncorrected boundary.
The first time point is the summation of the individual structures (after
applying boundary correction). Given that each structure
is segmented independently there are no constraints regarding overlap of
structures. Overlap is typically minimal and usually restricted
to the structure's boundary voxels. In the first time point voxels where
there was overlap will have an intensity equal to the sum of the labels
(these are not in any colour map or list).
Cheers,
Brian
> Dear FSLers and developers,
>
> Congratulations on your new release. The installation went fine with my
> machine so I decided to try out FIRST.
>
> I have general questions about its usage and the images created by
> FIRST. I ran “run_first_all im1 3 output_name” and my PC took about
> 2hrs to finish. I haven’t tried “first_utils” yet, so my questions are
> regarding the output image of first_all.
>
> PC specification: AMD64 1.8GHz 2GM RAM Centos 64
>
> #Q1 Would you recommend running bias-field correction before FIRST
> pipeline? I didn’t find any recommendation on this in the website, so I
> appreciate if you have any suggestion.
>
> #Q2 Can I consider the output image as an alternative to ROI ?
> After looking at the output image using FSL view, I thought I can create
> a binary mask in each stricture, and apply them onto other modality
> like ADC and FA as rough ROI. Is there something wrong with the
> concept here?
>
> #Q3 When total volume is computed in each structure, does it consider
> partial volume effect?
>
> #Q4 The image intensity in the output image does not seem to correspond
> sub- cortical label in the FSL website
> (http://www.fmrib.ox.ac.uk/fsl/first/cma_subcortical_label.html). It
> doesn’t seem to match with free-surfers’ colormap in “aseg” file either.
> Correct me if I’m wrong. I appreciate if there is any look up table
> for output intensity in the output image and anatomical labeling.
>
> With regards,
>
> Hedok
|