[Mristudio-users] corpus callsoum segmentation

[susumu mori]

Hi Sona,

This is a question about cross-subject brain registration. Your are
basically asking how we can identify the corresponding areas of the CC
across multiple subjects. Witelson scheme is one of them, which divide the
CC into the seven segments based on pre-defined distance-based criteria.
There are two often-counteracting factors in the mapping; reproducibility
(reliability) and granularity (the number of segmentation, or the amount of
localization information).

One extreme of the granularity is the entire CC, in which we get one number
for the entire CC (e.g. the size of the mid-sagittal CC). The reliability of
the CC definition is high across the subjects because we can unambiguously
define the mid-sagittal CC for any subjects, but the localization
information is the poorest. If only the genu of the CC is abnormal, such an
effect would be diluted by measuring the entire CC at once.

Another extreme is the pixel-based analysis, in which we have to make a
complete mapping of each pixel between the two subject. If there are 5,000
1x1x1 mm pixels within the mid-sagittal CC, we need to map the all 5,000
pixels to another brain. This is basically the same as making the two CC
shapes the identical by image transformation. This has the highest
localization information but the reliability is a tough issue. Depending on
which transformation algorithm you use, you get different results. This
approach has a potential to pin-point a small abnormal region within the CC,
but such an effect could be diluted if there is transformation error. Also,
each pixel is very noisy and a pixel-based statistics may have poor
sensitivity. Often you need to apply a filtering that introduces
pixel-averaging for nearby pixels.

The Witelson approach is somewhere between these two approaches. It has only
7 very large segments in which 100s of pixels are averaged.

Now your question is, if we provide a tool to perform the Witelson
segmentation. The short answer is no, but if you take your question as a
general brain registration question, the answer is yes.

By using DiffeoMap, you can transform one brain to the other. We are using a
very advanced diffeomorphic transformation developed by Michael Miller. So
you will find the transformation results are of quite high quality. Once you
transform one brain to the other, you can do pixel-based analysis.

Then you can move the transformed images into RoiEditor, in which you can
apply a pre-segmented brain atlas to segment the entire brain up to 176
areas. Currently, the CC is segmented to three areas: genu, body, and
splenium. In this atlas, you can create your own segment. For example, you
can segment the atlas CC to 7 Witelson segments. Then transform the atlas to
each subject (or transform each subject to the atlas) to apply these
segments to registered brain.

So, I would recommend you to take a look at www.mristudio.org and find what
you can do with DiffeoMap and RoiEditor.

Susumu

On Mon, Feb 1, 2010 at 11:58 AM, Sona Saksena <saksena.sona at gmail.com>wrote:

> Hello,
>
> I would like to calculate ADC and FA values in the seven segments of corpus
> callosum divided according to the Witelson scheme. CAn I do that using
> DTIstudio software. Please let me know how to divide the corpus callsoum
> into seven segments using DTIstudio software or suggest any other possible
> way to do it.
>
> Any help will be highly appreciated.
>
> thanks
>
> regards
>
> Sona Saksena
>
>
> _______________________________________________
> Mristudio-users mailing list
> Mristudio-users at mristudio.org
> http://lists.mristudio.org/mailman/listinfo/
> Unsubscribe, send a blank email to:
> Mristudio-users-unsubscribe at mristudio.org
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mristudio.org/pipermail/mristudio-users/attachments/20100201/a3886a0a/attachment.html