Hi Richard,
On Jul 1, 2013, at 6:21 PM, Richard Abrich wrote:
> Hi Luca,
>
> I managed to compile the latest source from the github repo on Windows 7
> 64-bit. However, I was getting the following error in
> vmtkLevelSetSegmentation.py on line 314:
>
> "AttributeError: 'module' object has no attribute
> 'vmtkImageInitialization'"
>
> For whatever reason, "import vmtkscripts" at line 22 wasn't pulling in
> vmtkImageInitialization. I verified this by dumping the object's properties
> right before line 314:
>
> import vmtkimageinitialization
> l = dir(vmtkscripts)
> self.PrintLog('\n'.join(l))
>
> The only module missing is vmtkImageInitialization.
>
> I got around this by replacing line 314:
>
> self.vmtkImageInitialization = vmtkscripts.vmtkImageInitialization()
>
> with this:
>
> import vmtkimageinitialization
> self.vmtkImageInitialization =
> vmtkimageinitialization.vmtkImageInitialization()
Thanks for looking into this, I'll look into why vmtkscripts does not include
vmtkimageinitialization, from a first check it's not clear to me.
> I then call the script as follows:
>
> vmtkimagereader -f dicom -d C:/DICOM/ --pipe
> vmtkimageinitialization -interactive 0 -method collidingfronts -sourcepoints
> 210 214 208 196 196 200 -targetpoints 204 204 204 188 190 193 --pipe
> vmtklevelsetsegmentation -iterations 100 -advection 1.0 -curvature 0.1
> -propagation 0.0 --pipe vmtkimageviewer -i @.LevelSets
>
> I currently have two questions:
>
> 1) vmtkLevelSetSegmentation does not seem to have a non-interactive mode, as
> the above call brings up the VTK window and asks for the level set
> segmentation parameters. This is also apparent in the source code. Luca, do
> you have an uncommitted version of this file that supports non-interactive
> mode?
It should work, though. At line 286, if NumberOfIterations is set the script
just runs level sets starting from InitialLevelSets, merges the result on the
output and exits. Maybe send me the text output that gets printed while the
pipe runs, it will help me understand what's going on.
> 2) The pipe to vmtkImageViewer at the end of the command does not display the
> segmentation, but the DICOM volume. How can I pipe the segmentation instead?
This way:
... --pipe vmtkimageviewer -i @.o
in this case you need to pipe the level sets explicitly because the associated
variable name is LevelSet, while image viewer expects Image. The first upstream
variable with that name is the output of the reader, which is what gets matched
in your case (but probably the explanation is superfluous because I saw you
were already doing an explicit piping there). The problem is that pipe syntax
refers to the option name (o in this case) while you are using the variable
name (this is how it works, although the variable name would make sense too).
Best,
Luca
>
> Thanks,
> Richard
>
>
>
>
> On Wed, Jun 26, 2013 at 5:53 PM, Richard Abrich <abri...@gmail.com> wrote:
> Answering my own question again:
>
> It turns out that the Colliding Fronts initialization code didn't support
> multiple start and end points. Lines 295-296:
>
>
> seedIds1.InsertNextId(self.Image.ComputePointId([self.SourcePoints[0],self.SourcePoints[1],self.SourcePoints[2]]))
>
> seedIds2.InsertNextId(self.Image.ComputePointId([self.TargetPoints[0],self.TargetPoints[1],self.TargetPoints[2]]))
>
> Need to be replaced with:
>
> for i in range(len(self.SourcePoints)/3):
>
> seedIds1.InsertNextId(self.Image.ComputePointId([self.SourcePoints[3*i+0],self.SourcePoints[3*i+1],self.SourcePoints[3*i+2]]))
>
> for i in range(len(self.TargetPoints)/3):
>
> seedIds2.InsertNextId(self.Image.ComputePointId([self.TargetPoints[3*i+0],self.TargetPoints[3*i+1],self.TargetPoints[3*i+2]]))
>
>
> Now it works.
>
>
>
> On Wed, Jun 26, 2013 at 5:12 PM, Richard Abrich <abri...@gmail.com> wrote:
> Sorry for the deluge of emails. I wasn't able to successfully get the latest
> version of the VMTK source to run properly, so instead I just downloaded the
> latest version of vmtkImageInitialization.py, which accepts the parameters
> you specified.
>
> My question is this: the description for the SourcePoints parameter is "list
> of source point IJK coordinates". Am I able to specify more than one
> SourcePoint and TargetPoint, and have it iterate over them? The alternative
> would be to save the results to disk after each call and then merge them all
> at the end before performing the LevelSetSegmentation, which seems pretty
> hacky.
>
>
>
> On Wed, Jun 26, 2013 at 11:07 AM, Richard Abrich <abri...@gmail.com> wrote:
> Hi again,
>
> I noticed that the latest version of VMTK in the Github repo has those
> parameters you specified. I'm currently compiling it, and will let you know
> how it goes.
>
> Thanks,
> Richard
>
>
>
> On Tue, Jun 25, 2013 at 5:37 PM, Richard Abrich <abri...@gmail.com> wrote:
> Hi Luca,
>
> Sorry I forgot to CC my last email to the mailing list.
>
> A couple of addendums to my previous email (below):
>
> 1) I converted my data to the .mha format and am using the exact command you
> suggested.
>
> 2) I used vmtk-exe.pyc instead of vmtk-exe.pyw in order to view the output.
> The relevant lines seem to be:
>
> Parsing options vmtkimageinitialization
> vmtkimageinitialization error: unknown option -interactive
>
> Just FYI, here is the revision information in the vmtkimageinitialization.py
> script:
>
> ## Date: $Date: 2006/05/31 10:51:21 $
> ## Version: $Revision: 1.19 $
>
> Is there a newer version I can grab from somewhere?
>
>
> On Tue, Jun 25, 2013 at 5:13 PM, Richard Abrich <abri...@gmail.com> wrote:
> Hi Luca,
>
> Thanks so much for your help. I have a couple of questions:
>
> 1) My data isn't .mha format, but rather stored as a dicom set. I tried to do
> something like this instead:
>
> vmtkimagereader -f dicom -d C:/DICOM --pipe vmtkimageinitialization
> -interactive 0 -method collidingfronts -lowerthreshold 200.0 -sourcepoints 54
> 25 12 -targetpoints 89 60 24 --pipe vmtklevelsetsegmentation -iterations 200
> -advection 1.0 -curvature 0.1 -propagation 0.0 -ofile C:/out.vtk
>
> However, the output file wasn't written, and I'm not sure how to check
> whether anything happened.
>
> 2) On a related note, I notice that in your example, you specify the ifile
> parameter for vmtklevelsetsegmentation again, even though you are piping the
> results from vmtkimageinitialization. Is this necessary? How can I do this
> using dicom sets instead?
>
> 3) A few of the parameters you specify for vmtkimagereader (e.g. interactive,
> method, lowerthreshold, sourcepoints, targetpoints) aren't listed in the
> documentation at http://www.vmtk.org/VmtkScripts/vmtkimageinitialization/ . I
> can see that LowerThreshold and UpperThreshold are properties of the
> vmtkImageInitialization object in vmtkImageInitialization.py, but the other
> parameters are a mystery. Where can I read more about them?
>
> Thanks again,
> Richard
>
>
> On Thu, Jun 20, 2013 at 4:41 PM, Luca Antiga <luca.ant...@orobix.com> wrote:
> Hi Richard,
> indeed there is a way. The initialization phase in vmtklevelsetsegmentation
> is performed by vmtkimageinitialization,
> which accepts parameters at the command line and whose output can be piped to
> vmtklevelsetsegmentation (in which
> case the latter will use the initialized level sets generated by the former).
> vmtklevelsetsegmentation can also run
> non-interactively by specifying iterations and evolution parameters at the
> command line. Here's a sample pipe:
>
> vmtkimageinitialization -ifile your_image_file.mha -interactive 0 -method
> collidingfronts -lowerthreshold 200.0 -sourcepoints 54 25 12 -targetpoints 89
> 60 24 --pipe vmtklevelsetsegmentation -ifile you_image_file.mha -iterations
> 500 -advection 1.0 -curvature 0.1 -propagation 0.0 -ofile level_set_file.mha
>
> One caveat: sourcepoints and targetpoints are in IJK coordinates (i.e. voxel
> indices), not physical coordinates.
>
> Hope this helps,
>
>
> Luca
>
>
> On Jun 18, 2013, at 10:51 PM, Richard Abrich wrote:
>
>> Hello,
>>
>> I'm interested in calling vmtkLevelSetSegmentation from a script. I want to
>> run colliding fronts with arbitrary thresholds and start/end position
>> vectors for the initial segmentation, and then run the level set
>> segmentation with arbitrary parameters.
>>
>> Unfortunately, there doesn't seem to be any way to specify these parameters
>> from the command line. What might be the easiest way to accomplish this?
>>
>> Thanks,
>> Richard
>> ------------------------------------------------------------------------------
>> This SF.net email is sponsored by Windows:
>>
>> Build for Windows Store.
>>
>> http://p.sf.net/sfu/windows-dev2dev_______________________________________________
>> vmtk-users mailing list
>> vmtk-users@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/vmtk-users
>
>
>
>
>
>
>
------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:
Build for Windows Store.
http://p.sf.net/sfu/windows-dev2dev
_______________________________________________
vmtk-users mailing list
vmtk-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/vmtk-users
