So, depending on the parameter sets, I get different results. So how does one go about deciding these parameters?
Yes it is convenient to run SW, but if one has to test hundreds of different options and simply try to visualize the outcomes to decide the parameter set, that’s a daunting task and I would argue quite subjective.
How do you guys decide you find the optimal parameter set for a given dataset? Especially for large studies (say with hundreds of samples).
M
![]()
![]()
From: shapeworks-users-request@sci.utah.edu [mailto:shapeworks-users-request@sci.utah.edu]
On Behalf Of Praful Agrawal
Sent: Tuesday, August 30, 2016 11:45 AM
To: shapeworks-users@sci.utah.edu; Murat Maga
Cc: Shireen Elhabian
Subject: Re: [shapeworks-users] ShapeWorksRun crashing with an itk error
Sure, Good Luck!
On Aug 30, 2016, at 12:42 PM, Murat Maga <maga@uw.edu> wrote:
Yes, I am still testing. Points seems to converge better with lower ending regularization. I will try your suggestions as well. Thanks Praful…
Btw.. seems like the correspondence model is not optimal.. probably need higher relative weighting
Ohh.. set the number of clusters to 0 in reconstruction parameters.
On Aug 30, 2016, at 12:16 PM, Murat Maga <maga@uw.edu> wrote:
OK. Some more progress with SW but here is a weird situation this is what groomed surfaces look like after optimization
This is what they look like after reconstruction
Quite a bit of detail is now lost and there is plenty of warping. Reconstruction parameters are default in SW studio.
I need to look up a sample parameter file and check it on my end. I have a couple meetings till late afternoon, will try to send it to you by evening or tomorrow morning.
On Aug 30, 2016, at 10:31 AM, Murat Maga <maga@uw.edu> wrote:
Are there any instructions on how to construct the parameter file for reconstructions?
On spacing issue Shireen might have some experience and/or suggestions.
For visualization you should be able to use the final correspondence model generated from command line, the only thing is you will have to construct the parameter file. Then you should be able to do visualization and reconstruction in studio.
On Aug 29, 2016, at 11:19 PM, Murat Maga <maga@uw.edu> wrote:
So I made some progress but still have two major issues.
2. I
can’t visualize the results I got from the command line tool in ShapeWorksStudio. The shapeworks viewer that came with the command line tool does not work for me due to QT error. Is there any other alternative?
Any suggestion is much appreciated,
They seem to be similar but not exact same voxel spacing. The latest version of shapeworks requires exact same voxel spacing, previous version had inbuilt prep tools to resample the volumes and make them to common spacing.
On Aug 26, 2016, at 12:41 PM, Murat Maga <maga@uw.edu> wrote:
So I have a different problem now. I padded the images and they all look like this, which I think is sufficient padding.
Groom successfully completed on Linux with these parameters:
<?xml version="1.0" encoding="UTF-8"?>
<blur_sigma>0.1</blur_sigma>
<antialias_iterations>100</antialias_iterations>
<background>0</background>
<foreground>1</foreground>
ShapeWorksGroom ./shapeworks.groom.xml center isolate hole_fill antialias blur fastmarching
But then ShapeWorksRun fails with this command:
File: /root/Downloads/ITK/Modules/Core/Common/include/itkImageToImageFilter.hxx
Description: itk::ERROR: MaximumEntropyCorrespondenceSampler(0x15be380): Inputs do not occupy the same physical space!
InputImage Origin: [-7.2094060e+00, -9.7350384e+00, -5.6886340e+00], InputImage_1 Origin: [-7.1679193e+00, -9.7181870e+00, -5.7039522e+00]
InputImage Spacing: [9.6125413e-02, 9.7350384e-02, 9.4810566e-02], InputImage_1 Spacing: [9.5572258e-02, 9.7181870e-02, 9.5065870e-02]
When I bring the raw files into ShapeWorksStudio on Windows, it complains with this error. Yet as far as I can tell all images are of the same dimension and
size (see the output from c3d below).
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.4188 19.4701 11.3773]}; vox = [0.0961254, 0.0973504, 0.0948106]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3358 19.4364 11.4079]}; vox = [0.0955723, 0.0971819, 0.0950659]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3277 19.4718 11.4256]}; vox = [0.0955178, 0.0973591, 0.0952136]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3358 19.4593 11.4392]}; vox = [0.0955723, 0.0972963, 0.0953266]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3839 19.4031 11.4214]}; vox = [0.095893, 0.0970155, 0.095178]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.4188 19.4579 11.4345]}; vox = [0.0961254, 0.0972894, 0.0952879]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.4275 19.3802 11.4441]}; vox = [0.0961834, 0.0969009, 0.0953678]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3633 19.4461 11.3909]}; vox = [0.0957555, 0.0972306, 0.0949242]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.4451 19.477 11.4043]}; vox = [0.0963004, 0.0973851, 0.0950356]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3839 19.4593 11.4392]}; vox = [0.095893, 0.0972963, 0.0953266]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3358 19.392 11.4214]}; vox = [0.0955723, 0.09696, 0.095178]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.4235 19.4461 11.4842]}; vox = [0.0961569, 0.0972306, 0.0957019]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3839 19.4146 11.4043]}; vox = [0.095893, 0.0970731, 0.0950356]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3703 19.4031 11.4525]}; vox = [0.0958018, 0.0970155, 0.0954372]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3767 19.4788 11.3831]}; vox = [0.0958447, 0.0973942, 0.0948596]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.4451 19.4364 11.3909]}; vox = [0.0963004, 0.0971819, 0.0949242]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3189 19.3286 11.3219]}; vox = [0.0954591, 0.0966428, 0.0943491]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.4321 19.4472 11.4345]}; vox = [0.0962137, 0.097236, 0.0952879]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3563 19.4828 11.3801]}; vox = [0.0957089, 0.0974138, 0.0948343]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3767 19.4593 11.4174]}; vox = [0.0958447, 0.0972963, 0.0951446]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.4275 19.3567 11.3976]}; vox = [0.0961834, 0.0967834, 0.0949802]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3421 19.3567 11.4118]}; vox = [0.0956142, 0.0967834, 0.0950982]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.4321 19.4143 11.4392]}; vox = [0.0962137, 0.0970716, 0.0953266]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3899 19.5094 11.4261]}; vox = [0.0959324, 0.0975468, 0.0952173]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.4106 19.4807 11.3909]}; vox = [0.0960705, 0.0974037, 0.0949242]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.4106 19.414 11.4043]}; vox = [0.0960705, 0.0970702, 0.0950356]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3698 19.5166 11.4907]}; vox = [0.095799, 0.0975828, 0.0957555]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3703 19.4031 11.4655]}; vox = [0.0958018, 0.0970155, 0.0955455]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3776 19.3802 11.4842]}; vox = [0.0958506, 0.0969009, 0.0957019]; range = [0, 1]; orient = RAI
Image #1: dim = [150, 200, 120]; bb = {[0 0 0], [14.3421 19.4031 11.3976]}; vox = [0.0956142, 0.0970155, 0.0949802]; range = [0, 1]; orient = RAI
On Aug 26, 2016, at 10:34 AM, Murat Maga <maga@uw.edu> wrote:
So what exactly is the padding issue? Too much padding, too little?
This seems like a padding issue as the new points generated with random seeds seem to be falling outside image domain.
On Aug 26, 2016, at 1:28 AM, Murat Maga <maga@uw.edu> wrote:
Any ideas what might be causing this error (log file is attached, but there is not much else in it)
....terminate called after throwing an instance of 'itk::ExceptionObject'
what(): /root/Downloads/shapeworks/src/ITKParticleSystem/itkPowerOfTwoPointTree.txx:125:
itk::ERROR: PowerOfTwoPointTree(0x4936160): Point [6.79247, -1.12035, -1.23879] is not contained within tree domain [-5.8547, -11.3321, -5.84979] - [5.8547, 11.3321, 5.84979]
./runSWrun.sh: line 2: 40528 Aborted ShapeWorksRun shapeworks.run.xml
|
