[Insight-users] Guidelines for Bspline mesh sizes.

[Matt McCormick]

Hi Tim,

Some guidelines can be found here:

  http://slicer.org/slicerWiki/index.php/FAQ#What.27s_the_BSpline_Grid_Size.3F

Yes, in general, the number of grid points is not too dependent on the
pixel density.  However, if there are only a few slices in the third
direction, for example, then the grid should be much more sparse.  An
overly dense grid will overfit the motion, but if there are
insufficient voxels, the motion may be undersampled.  Increasing the
grid density in a multi-resolution scheme is a good approach.

HTH,
Matt

On Tue, Feb 4, 2014 at 9:37 AM, Tim Allman <dr.tim.allman at gmail.com> wrote:
> The examples, tests and ITK Software Guide have little to say about the size
> of the Bspline mesh to use in deformable registration and the literature
> seems to be all over the place. In general, for a 2D registration the number
> of nodes in each dimension is usually the same even if the number of pixels
> or voxels in each dimension is quite different.
>
> Does anyone have any comments about a rational way to set up the mesh? I
> would like especially some guidance with multilevel registrations and 3D
> registrations of images with only a few slices.
>
> Thanks,
>
> Tim
>
> --
> Tim Allman, Ph.D.
> 35 Margaret Street,
> Guelph Ontario N1E 5R6
> Canada
>
>
> _____________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Kitware offers ITK Training Courses, for more information visit:
> http://www.kitware.com/products/protraining.php
>
> Please keep messages on-topic and check the ITK FAQ at:
> http://www.itk.org/Wiki/ITK_FAQ
>
> Follow this link to subscribe/unsubscribe:
> http://www.itk.org/mailman/listinfo/insight-users
>