[Mristudio-users] Counterintuitive LDDMM results

[susumu mori]

Dorian,

I assume that you are talking about the alpha value you specified when you
submitted LDDMM and you used 0.005 and 0.004.

If you put two numbers (you can put up to three numbers), we do two
CONSECUTIVE LDDMM; first using alpha=0.005 and then alpha=0.004. The
smaller the number the more elastic.

In this case, because most of the image matching is already accomplished by
the first LDDMM using 0.005, the second LDDMM with 0.004 did very small
amount of transformation, improving local mismatch.

To see the results with 0.004, you first have to combine two transformation
matrices, 0.005 and 0.004, and then do a single transformation. If you add
0.005 and 0.004 transformation matrices, you get the same matrix as the
"overall" transformation matrix you also got.

So, you rarely use the 0.004 matrix alone.

If you submit LDDMM twice independently; one with 0.005 and the other with
0.004, then you can compare two results with different elasticity values,
although 0.005 and 0.004 are two close to see large differences.

For young adult brains without atrophy, you can use 0.005. When atrophy is
severe, we recommend three alpha values with 0.01/0.005/0.002.


On Wed, Mar 13, 2013 at 4:28 PM, Dorian P. <alb.net at gmail.com> wrote:

> Hi all,
>
> I sent this question yesterday, not sure if it went through because of the
> attachment.
>
> I did a single channel LDDMM on a subject with two thresholds: 0.005 and
> 0.004. Suprisingly, the lower threshold (0.004) doesn't change much the
> brain of the subject, preserving the callosum shape. The other threshold
> (0.005) has worked better on reshaping the subjects callosum to the
> template shape. This is counterintuitive to say the least, because one
> would expect to have more morphing at 0.004 than with 0.005.
>
> Anybody knows why?
>
> Thank you
>
> Dorian
>
> P.s. Picture is here: http://tinypic.com/r/21ogmf/6
>
> _______________________________________________
> 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/20130313/282cdb04/attachment.html