[MINC-users] nlpfit deformation fields and mincblob

[Jonathan LAU]

Hi Marc,

When applying mincblob -det to the grid file (produced by nlpfit):
>
> what exactly is the result: the determinant of the jacobian of deformation
> field?


Yes.  In case you are interested, there are other operations available in
mincblob (use -help for details).

What is the meaning of a negative value?


The determinant can be a value between -1 and 1.  A negative value might be
interpreted as a localized decrease in volume, but be careful of this
interpretation on a raw output grid file as it will probably be quite noisy.

How can one get the log of
> the jacobian?


In the instance I've seen, it is applied as the following given
input_det.mnc
mincmath -const 1 -add input_det.mnc output_det_plus1.mnc
mincmath -log output_det_plus1.mnc output_log_det.mnc

I believe that you add 1 in order to get the range to be >0, but I wonder if
there's a better theoretical reason for this.

is the result in source or target space?


Not too familiar with nlpfit, but I believe that the nonlinear grid would
take you from source to target space.

does it make sense to resample the result?


Not sure what you mean by this?  The output_log_det.mnc file is in source
space though.

Hope this helps!

jon