Folded sample

[Maia R.]

Dear all,

I'd like to know why some samples from my distal femur model are folded as in the picture ?

I used a multiscale kernel in the registration step and I have a very small dataset (20).

Thank you very much,

Best regards

[Max]

Hi Maia,

I guess folds during sampling can occur if the mean shape already has (i) some small overhangs and/or (ii) small-scale, non-smooth structures. Another possible explanation could be that some of your shape model parameters are chosen too big or too many of them are at the very end of the allowed range (plus/minus three times the standard deviation).

Best

Max

[V R]

Hello, 

I have 400 coefficients from a multiscale kernel. How can I filter them so as to keep the most intersesting ones ?

Thanks

--
You received this message because you are subscribed to the Google Groups "scalismo" group.
To unsubscribe from this group and stop receiving emails from it, send an email to scalismo+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/scalismo/b71c23d9-93a1-4c71-bd80-a04fb62d303bn%40googlegroups.com.

[Andreas Morel-Forster]

In scalismo the rank of the model can be reduces. This will keep those modes of varitions whith the most variance (this is most often that what you might refere to as "interesting").

A statistical shape model in scalismo offers for example the method truncate(newRank), which reduces the model to only keep the first 'newRank' number of basis functions / coefficients. This can be seen also as some kind of regularization.

Best, Andreas

To view this discussion on the web visit https://groups.google.com/d/msgid/scalismo/CAL8G77%2BqsuNfVNAyP3hD3Oe8F3xZ0cBLNQTXcOMCb5Q_nX%2B55w%40mail.gmail.com.

[V R]

Thank you very much.

Best regards.

To view this discussion on the web visit https://groups.google.com/d/msgid/scalismo/7b7fa2d7-bf17-1d9a-63d7-9e55ccc9c83e%40gmail.com.