Applicability to other problems:
It has a broad range of applications beyond the two examples given.

Suggestions for future work:
None

Requests for additional information from authors:
None

Additional Comments:
None

Summary:
This work presents an extension to VTK classes implementing new functionalities that are shown in the applications MedINRIA and CardioViz3D.
Hypothesis:
NA
Evidence:
NA
Open Science:
This work is a nice example of the Open Science. Although the applications (MedInRIA, CardioViz3D) are not totally open source, the extension presented here is the kernel of these applications and it is available. Also, examples and data are open to test the contribution.
Reproducibility:
I tried to reproduce the author's results but I wasn't able due to I didn't get to compile the code under linux. I got an error that says that it doesn't find the library /usr/lib64/libfreetype.so. I have this library in the directory /usr/lib, because I'm not using a 64-bits platform. If I do a symbolic link to allow that the code find this library, I got a similar error with another library (libtiff.so) that is in the same directory. I will try to check why I'm getting this error, but till this review I haven't got to compile it :-(.
Use of Open Source Software:
The work is based on open source packages like VTK and ITK, and it is a good contribution for the VTK's community.
Open Source Contributions:
[Do the author's provide their source code? Is it in a form that is usable? Do they describe clearly how to use of the code? How long did it take you to use that code?]
Authors provide the source code and some examples to run it. The problem is that I haven't got to compile it under linux, although I've seen that other reviewers has got it under windows. I have spent two/three hours trying to compile it, but finally I haven't got.
Code Quality:
The code is clear and follows the VTK's style. It is easy to understand for VTK developers and users.
Applicability to other problems:
This work is useful for multi-dimensional or multi-valued dataset visualization in general, not only in medical image area.
Suggestions for future work:
Perhaps to include a precompiled version in its web site.
Additional Comments:
Nice work and a good contribution.

Summary:
Same as rev. 1

Hypothesis:
Same as rev. 1

Evidence:
Same as rev. 1

Open Science:
Same as rev. 1

Reproducibility:

I downloaded the code and compiled it with the same options as revision 1. Note: the source tarball include the svn revision information (.svn directories). It would be better to use the "export" option of svn instead of the "checkout" option when making the tarball.

 1./ Under windows XP 64, MSVC 8 2005, CMake 2.4.6, vtk 5.0.2, ITK 3.2.0. Everything compiled for win32 (not 64bits). 

- Without options, it now compiles. It seems that the ITK dependency has been removed (I did not check but I trust the authors did not hardcoded it :D). The same examples are available and running except of course the main application (INRIAviz4D).

-USE_ITK on
Same as rev. 1

2./ Under windows XP 64, MSVC 8 2005, CMake 2.4.6, vtk from CVS with Tcl support, ITK 3.2.0. KWWidget from CVS. Everything compiled for win32 (not 64bits).
Same as rev. 1.

3./  Under windows XP 64, MSVC 8 2005, CMake 2.4.6, vtk 5.0.2, ITK 3.2.0., wxWindow 2.8.4 from installer. Everything compiled for win32 (not 64bits).
It now compiles.

Use of Open Source Software:
Same as rev. 1

Open Source Contributions:
Same as rev. 1 

Code Quality:
Same as rev. 1

Applicability to other problems:
Same as rev. 1

Suggestions for future work:
There is a dashboard (which is good) but there is no test available. It would be nice to have a test suite, and also t o have in the source code the minimum amount of data to run examples and tests. The authors reported this as work in progress.

As for the tensor visualisation, it could be nice to offset the glyphs so that they can be visible from both sides of the orthogonal plans. Right now, they are only visible from one side. Still not addressed in rev. 2. I don;t know if the authors are going to address this or not.

Requests for additional information from authors:
-

Additional Comments:
The biggest issues (itk dependency, window text, compilation with wx) have been very quickly addressed by the authors. There is now a tarball associated with the revision of the paper, which makes it easier to review (same code between two reviewers) as the svn server is very active. Only a few enhancements are proposed here. As for the features, this deserves a 5 stars. If the code was ever to be ported to VTK, there would be an effort to be done on the test side, but that would be it.

Summary:
This paper address several problems that are generic to anybody working on 3D(potentially +t) Images processing, namely the synchronization between differents views of the same object, the temporal aspect of some datasets (cardiac MRI), and the common visualization of different types of data (images, volume/surface meshes, subsets of each, ...).

Hypothesis:
- The user wants to visualize a dataset from different views (standard case for 3D images) and want the modification made on one view's settings to be reflected on the other views of the same object.

- The user wants to see in a common view an image and a mesh (think about the result of an isosurface extraction, of a segmentation, ...)

- The user has a 3D+t dataset (possibly polydata+t).

Evidence:
There is no hard evidence , as there is no real hypothesis either. The cases the authors are supporting with their code are common enough (less common maybe for the 3D+t but still of interest) to justify the integration in the toolkit. 

Open Science:
Developped entirely on top of VTK (even if the provided code uses a little bit of ITK, KWWidget/wxWindow for the examples), the code is provided under BSD licence. It's  thus available to whoever wants to re-use it. The data need to be downloaded separately.

Reproducibility:

1./ Under windows XP 64, MSVC 8 2005, CMake 2.4.6, vtk 5.0.2, ITK 3.2.0. Everything compiled for win32 (not 64bits). 

- Without options, the code simply does not compile. It seems that ITK is part of the core, and not optional as stated in the paper (page 3, "ITK 3.2.0 (for some optional components. In that case, those classe can not be integrated in VTK without adding a dependency on ITK, which, I guess, is not allowed. If the ITK dependency was not removed, this code should belong to ITK Applications ( I don't see any other NA-MIC repository suitable). Note: the vtkItk repository contains code that was apparently taken over from InsightApplications/Auxiliary/vtk. This code deals with tranfering Images between ITK and VTK.

-USE_ITK on
The code compiles. Of the 7 tests availables only two are discussed in the paper. CompareImageManager, DataManager (paper: A.2), FibersManager, INRIAviz4DLauncher, IsosurfacesManager, SynchronizedViews (paper: A.1), TensorManager.
INRIAviz4DLauncher did not seem to have any visible effect.
CompareImageManager illustrates a 3D checkboard with orthogonal visualization.
All the other *Manager are (impressive) examples of mixed visualization between images and UGrid/PolyData, where the UGrid/Polydata represent Fibers, Surfaces, Tensors, ...
The examples all ran fine, and reproduced the paper's statements.

2./ Under windows XP 64, MSVC 8 2005, CMake 2.4.6, vtk from CVS with Tcl support, ITK 3.2.0. KWWidget from CVS. Everything compiled for win32 (not 64bits).

Same remark for ITK.
It seems that the following line fails to recognize VTK  cvs:
vtkViewImage3D.cxx(231): #if VTK_MAJOR_VERSION==5 && VTK_MINOR_REVISION>=1.
It works fine with vtk 5.0.2.

Under this configuration, a new .exe (INRIAviz4D.exe) is created and all the other executable do not print an help message anymore when run with no arguments. the Launcher.exe now launch a global application which seems to wrap all the previous features within a nice KWWidget based API. Nothing about it in the paper, but it seems to illustartes some other points like the cropping/projections of UGrids/PolyData. I did not tested everything, which woul dhave been to long in the absence of documentation/tutorial.

 3./  Under windows XP 64, MSVC 8 2005, CMake 2.4.6, vtk 5.0.2, ITK 3.2.0., wxWindow 2.8.4 from installer. Everything compiled for win32 (not 64bits).

the correct wx directories were not found. I don;t know if it's due to this version of wx, of if it's because the authors expect an install for source.

Use of Open Source Software:
The author used VTK (/ITK/KWWidget...). They installed a DART server  to check compilation.

Open Source Contributions:
Do the author's provide their source code? Yes, on their own svn server, under BSD.
Is it in a form that is usable? Up to the problems stated above, yes.
Do they describe clearly how to use of the code? Yes, except for the INRIAviz4D interface.
How long did it take you to use that code? Half a day to use. I cannot evaluate how long it would take to cod eon top of it, but I can tell it's gonna be way faster than trying to code the equivalent feature based solely on VTK. Once again, that's the kind of fonctionality that everyone wants and thus recode several times.

Code Quality:
Source code is clean. Somme parts are left commented, but it's minor. The .h are commented and follow the dossygen style. Doxygen documentation is part of the CMake file. The overall quality is very good.

Applicability to other problems:
Any problem that requires visualization of 3D(+t) images, possibly with Meshes.

Suggestions for future work:
Is it possible to make the ITK part really optional? Right now, the ITK dependency prevents this code for being accepted in VTK.

There is a dashboard (which is good) but there is no test available. It would be nice to have a test suite, and also t o have in the source code the minimum amount of data to run examples and tests.  

The :"Powered by vtkINRIA3D" might be removed from the examples. I say might, because this is just examples, so it shoudl not impact other people's code. In any case, I would recommend removing the  "Powered by MedINRIA" in /vtkRenderingAddOn/vtkViewImage.cxx, which seems to belong to the core of vtkINRIA3D.

As for the tensor visualisation, it could be nice to offset the glyphs so that they can be visible from both sides of the orthogonal plans. Right now, they are only visible from one side.

Requests for additional information from authors:
Is there a tutorial of INRIAviz4D somewhere?
What is the expected configuration to build vtkINRIA3D with wx?

Additional Comments:
I did not test the cropping/projection of the meshes, but it seems to make a lot of sense and that's the first time I see such a feature.

I loved the synchronized view exemple. It is nioce to have each view interactor control one parameter (position, zomming, intensity, ... You might want to disable the interaction with the position in the 2D screen space though on the first windows. As one could just drag and move the image out of the windows (the ImageViewer1|2 classes do that I guess in vtk). It's really minor though. Great exemple.

I only give 4 stars to this because, although it's a great feature I would love to have integrated in VTK, there is a dependency over ITK, and also because parts of the code are not documented / not working out of the box. Would those problems be addressed, I would of course improve the rating.

Summary:
The authors address a key technical problem for researches in the field of medical image processing. A software tool that provides a simple interface for 3d visualization of datasets and algorithms results. Their approach is an extension to the well known visualization library - VTK with a very simple interface.

Hypothesis:
Non Applicable

Evidence:
The authors made their code open-source so any one can download and use it. However the paper just describe their toolkit, and do not made experiment checking how easy it is to implement some tool using their toolkit vs other toolkits. However it seems that their rationality is better than other toolkits for the specific task of development visualization tools for analyzing medical images.

Open Science:

The authors made their code available to download. Their code include testing software, Applications which build upon the described toolkit, and other examples. Images to using with in the example applications. The toolkit built using standard configuration tools in the community such as cmake, so the using of the code should be easy.

Reproducibility:
I download author's code. Configure it using cmake, and try to build it. However when i tried to Try to compile the compiler failed due to problem in the Makefile generated by the cmake program. So i did not able to run it.

I got an error said: "target use/lib required by SynchronizedViews did not found."

I suggest the authors explain exactly in their text or in the documentation of the code how to build it.

Use of Open Source Software:
The authors extend a very well known open-source library - VTK. this is relatively important advantage. Many users of ITK and VTK can use their code for building visualization applications.

Open Source Contributions:

The authors provide their code. The paper includes several examples of using the code. Although no building instructions are available, the building process is very similar to the building of ITK and VTK so it is very easy to build it..

Code Quality:
Their code seems to be cross-platform, but i did not check it. It seems that their code is high-quality, and can be extend in a very easy manner.

Applicability to other problems:
As I mentioned before, many researches in the field of medical image processing need software tools for interactive visualization of 3D datasets in order to evaluate their algorithms. Until now I did not find a software library tool that provide a simple interface for programming such tools. vtkinria3d do provide such an interface and I hope that when i succeed to compile it, I'll found it a very useful tool.

Suggestions for future work:
Fix the code so we will able to compile it.

Requests for additional information from authors:
Fix the compilation error I reported before.

Additional Comments: