Insight Journal - Diffeomorphic Demons Using ITK's Finite Difference Solver Hierarchy

Diffeomorphic Demons Using ITK's Finite Difference Solver Hierarchy

Vercauteren T., Pennec X., Perchant A., Ayache N.

Asclepios Research Group, INRIA Sophia Antipolis and Mauna Kea Technologies

Please use this identifier to cite or link to this publication: http://hdl.handle.net/1926/510

Published in The Insight Journal - 2007 MICCAI Open Science Workshop.

Submitted by Tom Vercauteren on 06-06-2007.

This article provides an implementation of our non-parametric diffeomorphic image registration algorithm generalizing Thirion's demons algorithm. Within the Insight Toolkit (ITK), the demons algorithm is implemented as part of the finite difference solver framework. We show that this framework can be extended to handle diffeomorphic transformations. The source code is composed of a set of reusable ITK filters and classes. In addition to an overview of our implementation, we provide a small example program that allows the user to compare the different variants of the demons algorithm.

Code

Automatic Testing Results by Insight-Journal Dashboard on Fri Jun 19 12:43:57 2009 for revision #11

expertise: 5 sensitivity: 5

This project passed all of its tests.
Click here for more details.

Go here to access the main testing dashboard.

Automatic Testing Results by Insight-Journal Dashboard on Mon Dec 1 05:46:34 2008 for revision #10

expertise: 5 sensitivity: 5

Automatic Testing Results by Insight-Journal Dashboard on Fri Nov 28 14:12:45 2008 for revision #9

expertise: 5 sensitivity: 5

Automatic Testing Results by Insight-Journal Dashboard on Fri Nov 28 14:12:36 2008 for revision #8

expertise: 5 sensitivity: 5

Automatic Testing Results by Insight-Journal Dashboard on Mon Mar 2 12:06:12 2009 for revision #7

expertise: 5 sensitivity: 5

Automatic Testing Results by Insight-Journal Dashboard on Fri Sep 21 08:02:14 2007 for revision #6

expertise: 5 sensitivity: 5

Automatic Testing Results by Insight-Journal Dashboard on Thu Sep 13 11:34:33 2007 for revision #5

expertise: 5 sensitivity: 5

Automatic Testing Results by Insight-Journal Dashboard on Thu Sep 13 11:29:12 2007 for revision #4

expertise: 5 sensitivity: 4

Automatic Testing Results by Insight-Journal Dashboard on Thu Jun 7 10:24:35 2007 for revision #3

expertise: 5 sensitivity: 5

Automatic Testing Results by Insight-Journal Dashboard on Thu Jun 7 10:16:17 2007 for revision #2

expertise: 5 sensitivity: 5

Automatic Testing Results by Insight-Journal Dashboard on Wed Jun 6 08:38:07 2007 for revision #1

expertise: 5 sensitivity: 5

Reviews

MICCAI workshop review by Kevin Cleary on 09-11-2007 for revision #4

expertise: 2 sensitivity: 4

Summary:
In this paper a modified version of Thirion’s Demons Registration algorithm is presented.As authors mention this is a companion paper to the paper “Non-parametric diffeomorphic image registration with the demons algorithm” published in MICCAI 2007 with the purpose of sharing the source code with the community. Instead of presenting in detail the algorithm they insist more on the ITK implementation issues.To understand the method and to see the evaluation of the algorithm the reader has to read also the initial paper.

Hypothesis:
Lie group structure on diffeomorphic transformations can be used in combination with some optimization tools on Lie groups to derive our diffeomorphic image registration algorithm.

Evidence:
They provide the source code with only one 128x128 two examples 2D example.I couldn’t execute the code on Visual Studio 8.0 because the file getopt.h was missing.

Open Science:
They provide the full source code based on ITK and they provide one data set.The actual paper contain no images only general description of the algorithm and details on ITK implementation.The original MICCAI paper contains images and results based on different data that is not available with this source code.

Reproducibility:
Did not run on Microsoft Visual Studio 8.0

Use of Open Source Software:
Their implementation is based on ITK.Actualy they are also one of the important contributors of the Demons registration in ITK.This algorithm is only an extension to the current ITK framework
Open Source Contributions:
The code is well written and it contains reasonable comments inside the code
Code Quality:
Seems it doesn’t compile on Visual Studio 8.0 because of an utility file: getopt.h

The files are following ITK coding style and it is easy to read

Applicability to other problems:

Suggestions for future work:

Requests for additional information from authors:

Additional Comments:

This review was written by Teo Popa in our laboratory and posted by Kevin Cleary.

Comment by Tom Vercauteren: Now compiles with VC++ 2003

Thank you for your review and for pointing out the fact that the code could not compile under VC++.

Compilation problems
=============
I had actually assumed that it would work under VC++ since I thought I wasn't using any linux specific code... I never realized getopt was not native on VC++.

I have now included a compatibility library for getopt to work on VC++. The code has now also been tested on Windows XP with VC++ 2003. Let me know if you still have problems using it.

Reproducibility of the MICCAI figures
=====================
You are indeed right to mention that we do not provide all the material to reproduce the figures presented in MICCAI. The reason is that the scripts we wrote are not clean enough for public reuse. The main goal of this submission was to make the algorithm available to the community and not the evaluation framework we use in the MICCAI paper. We would rather argue using a more principled benchmark such as the one submitted to this issue of the Insight Journal by Urschler et al.

Additional comments
============
I am not sure what you mean by "Actualy they are also one of the important contributors of the Demons registration in ITK". The only way my coauthors and I have contributed to the code in ITK before was through some bug reports.

Valuable contribution to enhance ITK\\\'s nonrigid registration capabilities by Torsten Rohlfing on 09-26-2007 for revision #3

expertise: 5 sensitivity: 5

Summary:
This submission provides an implementation of a diffeomorphic nonrigid demons registration algorithm. It comes with a stand-alone command line tool that provides many useful options and is fit for production use.

Evidence:
The authors provide test images.

Open Science:
Source code and test data are provided.

Reproducibility:
I have not tested the tool with the data provided by the authors. I have, however, run extensive tests on my own data in a longitudinal deformation-based morphometry study. I find the core algorithm to be very efficient and about as accurate as the original demons algorithm. I was also able to confirm that the produced transformations are essentially diffeomorphic. There occasionally appear to be small fractions of pixels with negative Jacobians, but this seems mostly connected to almost-folding input deformation fields that I sometimes provide to the algorithm.

Use of Open Source Software:
Uses ITK.

Open Source Contributions:
Source code is provided. Compiles successfully out of the box using current CVS version of ITK.

Code Quality:
The code quality is excellent. The code appears to conform with ITK coding style and is virtually fully documented. The bundled command line tool provides options for different update rules, multi-resolution registration, initial deformation fields, and intensity histogram normalization.

Applicability to other problems:
n/a

Suggestions for future work:
n/a

Requests for additional information from authors:
n/a

Additional Comments:
I would strongly advocate adding this contribution to ITK as soon as possible. The nonrigid registrations in the toolkit are a little outdated right now and could really use some newer algorithms. In particular diffeomorphic registration has become a hot topic recently, and it would be great to have a diffeomorphic registration algorithm available in ITK, in particular one that is as efficient as this one. I would also advocate including the command line tool that comes with the software in the ITK source tree.

Add a new review

Statistics


Global rating:
Review rating:	[review]
Code rating:
Views:	6187
Downloads:	1687

Send a message to the author

Information


Paper Id:	154
Categories:	Deformable registration, Generic Programming, Image, Mathematics, Optimization, Registration
Keywords:	Registration, Demons, Diffeomorphisms,
Toolkit:	ITK (moved into the sandbox), CMake
Revision:	11 (06-19-2009)
Status:	Open for public review

Data


Full download:	.zip
Paper:	view, .pdf
Source code :	Download

Share

Associated Publications


An ITK Implementation of the Symmetric Log-Domain Diffeomorphic Demons Algorithm
The Role of Open-Source Software in the Study of Embryogenesis
ITK-based Registration of Large Images from Light Microscopy: A Biomedical Application