Dear Charlotte,
The terms left and right, when used in FIRST apply to the real
left and right of the subject, and have nothing to do with the
display of the images. Hence if you ask for the left hippocampus
to be labeled then you will get the subject's left hippocampus,
regardless of which way around the data is stored and also,
independently, regardless of which way around a viewing tool
decides to display the data. You may see that the viewing tool
used shows things such that left structures appear on the right
of the image, or they may appear on the left. This is arbitrary,
although FSLView tries to show it in the "radiological" convention
with left displayed on the right side of the screen. However,
because this very, very quickly gets confusing, the way to be
sure what side is what on the display is to view the *labels* on
the image. You should see "L" and "R" attached to the image
view. These always correspond to the subject's left and right,
respectively. For the 3D viewer I believe that FSLView and all
other software that I'm aware of show the image as if it were a
real object that you were looking at. So if the anterior of the brain
is closer to you in that view then it will be like "radiological" but
if the posterior is closer (as if you were behind the subject) then
it will be like "neurological". In this case it is clearer as you can
tell from one view whether you are "in front" or "behind" the object
just as if it were a real subject in front of you.
With respect to your flipped case, the shape model that FIRST uses
has an asymmetry in it, based on the fact that there is significant
asymmetry in the population and hence the training data used in FIRST.
However, that being said, we normally find negligible asymmetry when
we do this test of flipping the same brain. So check that you are actually
flipping correctly and that you are comparing "like with like" in that you
compared the unflipped left with the flipped right (and not the flipped
left).
I hope this makes things clearer.
All the best,
Mark
On 17 May 2011, at 11:35, Charlotte Bernard wrote:
> Thanks. But I have others questions:
>
> 1/ if I have images in radiological convention: when I want to segment the left hippocampus, it corresponds in FIRST to the left or the right hippocampus?
> However, I suppose that for images in neuroradiological images, when I segment the left hippocampus, it corresponds in FIRST to the left hippocampus.
>
> 2/ I have a subject with a 3D scan fliped and the same 3D scan not flipped. When I perform FIRST, I don't obtain the same volume for left and right hippocampus. How is it possible?
>
> Thanks,
> Charlotte
>
>
>
> > Date: Mon, 16 May 2011 16:27:21 +0100
> > From: [log in to unmask]
> > Subject: Re: [FSL] FIRST and convention
> > To: [log in to unmask]
> >
> > Hi,
> >
> > Either convention is fine for FIRST, and all the other tools in FSL too.
> >
> > All the best,
> > Mark
> >
> > On 16 May 2011, at 15:47, SUBSCRIBE FSL Anonymous wrote:
> >
> > > Dear FSL user,
> > >
> > > I have a question about the convention of the analyzed images with FIRST. The images have to be in a radiological or neurological convention?
> > > Thanks,
> > > Charlotte
> > >
|