[Mristudio-users] MultiChannel LDDMM

[susumu mori]

Hi Jeff,

I think you can submit up to 4 or 5 channel LDDMM through DiffeoMap.

When you use more channels, the more memory it takes and could take longer
time for calculation. So you want to use the channels wisely. For example,
b0 and trace have very similar contrasts and won't improve the registration
that much. FA+Trace+b0 could put more weight on Trace/b0 contrasts but
other than that, the result should not be much different from FA+Trace. DWI
may provide some additional information and could improve the results. What
we often do is to submit 1/0 binary masking data for the ventricle. This
improves the registration a lot especially for patients with very enlarged
ventricles. This method, however, requires some additional work to define
ventricles.

Addition three channels of color images (red, green, and blue) is another
option. This theoretically allows us to match red fibers to red fibers,
green to green, and blue to blue, further improving the registration, but
there are a couple of things that require attentions. First, strictly
speaking, the fiber color can change due to transformation. Therefore, we
need to generate new color maps after each finite transformation. This
algorithm is not implemented in the current DiffeoMap. Second, red-to-red
matching sounds good for core white matter regions but for peripheral white
matter regions where R, G, B are mixed in a complex and in-reproducible
manner, you may not want to force red-to-red matching.

So,  it is not an easy task to find if addition of new contrasts really
improve registration in a reproducible way. It could also depend on the
amount and nature of the anatomical abnormality of your patient group.  So
I would encourage you to give some testing. Oh, one thing I want to add is,
if you use the color maps with the current DiffeoMap algorithm, you may get
some unhappy reviewers when you submit a paper, although, I believe one of
the original papers in this topic by C.F. Westin used this type of
simplified version of color-to-color registration.


On Mon, Jul 30, 2012 at 4:42 PM, Jeff Sadino <jsadino.queens at gmail.com>wrote:

> Hello,
>
> I am wondering why we only submit the FA and the Trace images.  It seems
> like we could submit the FA, Color, DWI, and Trace images.  I see that the
> information from these images come from the same "primary channel", but in
> my testing, the more images I submit, the better registration I get.  Is
> there anything wrong with submitting all four of these channels?
>
> Thank You,
> Jeff Sadino
>
> _______________________________________________
> 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/20120730/17030843/attachment-0001.html