Begin forwarded message:

From: Shireen Elhabian <shireen@sci.utah.edu>

Subject: Re: [shapeworks-users] ShapeWorks Questions

Date: February 5, 2018 at 8:06:02 AM MST

To: Katarina Ruehlicke <ruehlicke.k@husky.neu.edu>

Cc: Shireen Elhabian <shireen@sci.utah.edu>

Hi Katie,

It seems that you are not getting a good correspondence model. You might want to tune some parameters to get a better model (starting and ending regularization, relative weighting). Here is the link for the command line tools that I was talking about. 

https://github.com/SCIInstitute/shapeworks/releases

thanks and best regards

Shireen

------------------------------------------

Shireen Elhabian, M.Sc., Ph.D.
Research Assistant Professor

School of Computing
Scientific Computing and Imaging Institute
University of Utah
WEB 2815
72 Central Campus Drive, Salt Lake City, UT, 84112
Phone (801) 587-3206
Fax (801) 585-6513

Home Page: http://www.sci.utah.edu/~shireen

On Feb 2, 2018, at 10:25 AM, Katarina Ruehlicke <ruehlicke.k@husky.neu.edu> wrote:

Hi Shireen, 

Here, the first two snapshots are of the correct images that I can get if I do not mess around with the reconstruction portion. Following that, I changed some numbers in the reconstruction panel to produce the next two snapshots here, and as you can see, they are warped femurs. 

As of now, the only ShapeWorks that I have access to is the ShapeWorksStudio, so I'm not sure how exactly to proceed to find ShapeWorksView2. 

Thank you so much for your help and advice, 

Katie Ruehlicke 

Bioengineering Student

Northeastern University

ruehlicke.k@husky.neu.edu

<image.png>

<image.png>

<image.png>

On Thu, Feb 1, 2018 at 4:23 PM, Shireen Elhabian <shireen@sci.utah.edu> wrote:

Hi Katie,

If the reconstructed shapes are messed up (lots to twisting for example), this is an indicative of a bad correspondence model. How about the mean shape? Could you share a snapshot of the mean shape (with particles on it)?

Unfortunately, currently studio is not saving those reconstructed meshes. If you only want a dense mesh (based on the particle system) without preserving correspondence between those mesh, you might want to try out the shapeworks command line tool ShapeWorksView2. With this tool, you can export meshes for different samples (but they are not into correspondence).

best regards

Shireen

On 01/30/2018 03:20 PM, Katarina Ruehlicke wrote:

Hi Shireen, 

Thank you so much for your advice! I had just been experimenting with the number of clusters and was not entirely sure what it meant or if it impacted the MeshQC warning I was receiving. 

So I noticed that under my reconstruction panel, I only have the "Mesh Decimation %" option, instead of the max 1 you suggested. Either way, I tried to increase this number and the femurs are still appearing a little messed up compared to how they were before I reconstructed them. Any ideas?

Also one final thing, how am I able to save everything I have been working on as a .vtk file? When I try to save the project, the only option I'm given is to save it as a .xml file. 

Thank you so much, 

Katie Ruehlicke 

Bioengineering Student 

Northeastern University 

ruehlicke.k@husky.neu.edu

On Mon, Jan 29, 2018 at 8:47 PM, Shireen Elhabian <shireen@sci.utah.edu> wrote:

Hi Katie,

For getting more triangles (denser mesh), I would increase the “Mesh Decimation” (max 1) in the reconstruction panel.

For the number of clusters, is there a particular reason you want to have all samples in your dataset to be warped to the mean space? This is what happens when you use clusters = 0. I would set it to a small number (less than sqrt(N/2) where N is the number of samples).

The MeshQC warning might be fixed by trying out a different mesh decimation percentage. We are currently working on a more robust mesh decimation implementation that won’t produce this warning. In all cases, you will have a vtk mesh at the end but it will be denser if you get this warning.

Hope this helps.

thanks and best regards

Shireen

------------------------------------------

Shireen Elhabian, M.Sc., Ph.D.
Research Assistant Professor

School of Computing
Scientific Computing and Imaging Institute
University of Utah
WEB 2815
72 Central Campus Drive, Salt Lake City, UT, 84112
Phone (801) 587-3206
Fax (801) 585-6513

Home Page: http://www.sci.utah.edu/~shireen

On Jan 29, 2018, at 11:19 AM, Praful Agrawal <prafulag@cs.utah.edu> wrote:

Hi Shireen,

I think the query relates to surfworks output in ShapeWorksStudio. Kindly look into it.

Thanks,

Praful

On Jan 29, 2018, at 11:16 AM, Katarina Ruehlicke <ruehlicke.k@husky.neu.edu> wrote:

To whomever this may concern, 

I am currently trying to compare the femurs of lab mice, and would like to know if there is a way I can import all of the bones (or even just a sample of them) and then obtain an interpretation of the data and the differences between the femurs. As of now, I am importing .nrrd files, and would like to increase my mesh size / make the triangles smaller / etc. and then have a .vtk file to save. Any suggestions? I am also attempting to reduce the Cluster # to 0 instead of the 5 I have it set on, and it keeps producing a warning that "the MeshQC failed, but a dense mean was computed by VTK", and I have no idea what this means or how to fix it without restoring my defaults and running through the groom phase. Again, any suggestions would be much appreciated. 

Thanks, 

Katie Ruehlicke

Bioengineering Student 

Northeastern University

ruehlicke.k@husky.neu.edu