[ITK] Regularizing b-spline deformation in ITKv4

[Andriy Fedorov]

Nick,

we are comparing BRAINSFit in Slicer3 that was based on ITKv3 vs
BRAINSFit in Slicer4 based on ITKv4.

We have been using BRAINSFit/Slicer3 for a long time on a large number
of clinical research cases and we've been reasonably happy with the
results.

We need something that is integrated in Slicer, or can be implemented
in SimpleITK relatively easily. We need a tool that produces a
reasonable result within 2 minutes of computation, without the need to
change parameters depending on the dataset. So these are basically our
constraints. We were hoping we could improve BRAINSFit/Slicer4 by
adding a regularizer, or implement a subset of BRAINSFit functionality
in SimpleITK.

AF

On Wed, May 20, 2015 at 11:39 AM, Nicholas Tustison <ntustison at gmail.com> wrote:
> Hi Andriy,
>
> How are you comparing ITKv3 with ITKv4?  I assume you’re running
> ITKv3 through BRAINSFit but how are you running ITKv4?
>
> There are a couple of alternative B-spline methods if you’re interested.
> There’s B-spline SyN which is described here
>
> http://www.ncbi.nlm.nih.gov/pubmed/24409140
>
> and
>
> a Demon’s like B-spline algorithm described here
>
> http://www.ncbi.nlm.nih.gov/pubmed/19171516
>
> which both offer different regularization options that are available in
> ITKv4.  However, alternative explicit regularization models, such as
> thin plate splines, are not available in ITKv4.
>
> Nick
>
>
>
>
>
> On May 19, 2015, at 5:38 PM, Andriy Fedorov <fedorov at bwh.harvard.edu> wrote:
>
> Hi,
>
> we are in the process of evaluating the possibility of migrating
> workflows from ITKv3 to ITKv4 b-spline based registration (see details
> here [1]).
>
> One issue that is becoming apparent is that sometime the deformations
> we are getting with ITKv4 lack are a lot more irregular and
> unrealistic than what we used to get in ITKv3 (see case 9 on the page
> referenced).
>
> At the ITK community hangout last week it was suggested that there are
> some ways in ITKv4 to add regularization to b-spline transformation.
>
> Is there an example of b-spline regularization we could follow to see
> how this should be done? Is there a significant performance hit if
> this regularization is used?
>
> Thanks
>
> AF
>
> [1] http://www.na-mic.org/Wiki/index.php/BRAINSFit_prostate_registration
>
>
> The information in this e-mail is intended only for the person to whom it is
> addressed. If you believe this e-mail was sent to you in error and the
> e-mail
> contains patient information, please contact the Partners Compliance
> HelpLine at
> http://www.partners.org/complianceline . If the e-mail was sent to you in
> error
> but does not contain patient information, please contact the sender and
> properly
> dispose of the e-mail.
>
>