Dear all,

I have a question regarding the DTI output files of TumorSim1.2. I suppose the warped and modified tensors (after insertion of the tumor) are contained in the "xxx_modified_dti.mha" output file. As an example, I consider those files from Niethammer's database on Longitude with Pathology Challenge , named "Sim_XXX_modified_dti.mha" (XXX ranging from 071 to 094). The dimensions of those images are 256x256x1086. The diffusion information is indeed stacked along the z-dimension. When one scrolls from slice 1 to slice 1086, one encounters six volumes, which are supposed to be maps of the six tensor components (Dxx, Dxy, Dxz, Dyy, Dyz, Dzz respectively). However, the maps do not appear to give this information. Whereas the second and third volume indeed seem to represent off-diagonal components of the tensors (e.g. lower intensity), the fifth one does not look like an off-diagonal component. As a consequence, we appear to have 4 on-diagonal components and 2 off-diagonal components? I opened the files with different viewers and they all confirmed that there is something odd about the DTI output files.

Moreover, the tumor appears abnormally bright or abnormally dark in the DTI output files, depending on the tensor component under consideration. As an example, I included a slice from the Dxx volume of Sim_071_modified_dti.mha.

Has anyone encountered the same problems or has anyone been able to fit realistically looking tensors to the diffusion data after insertion of the tumor?

Thanks in advance,
Tom Haeck

Tom, thank you for letting us know about the problem with the DTI output. The tensors inside the tumor region were scaled inappropriately, which causes it to appear much brighter than the tensors outside the tumor.

I have released version 1.2.2 that addresses this issue, with minor improvements on the DTI computation. Please give it a try and let me know if there are still problems.

Dear Marcel
we tried to use Tumorsim 1.2.2 on Ubuntu 11.4 and Win7 but it dosen't seem to work.
On linux the problem is :
./tumorsim_1.2.2_linux64: /usr/lib/x86_64-linux-gnu/libstdc++.so.6: version 'GLIBCXX_3.4.15' not found (required by ./tumorsim_1.2.2_linux64)
./tumorsim_1.2.2_linux64: /lib/x86_64-linux-gnu/libc.so.6: version 'GLIBC_2.14' not found (required by ./tumorsim_1.2.2_linux64)
On win7 at fisrt it didn't find msvcp110.dll; when you add file .dll  it returns an unknow error.

Can you help me?

flavio

@Tom: try with case input 2 and 3 where the component Dyz seem to be correct.

Hi everyone,

When I tried to install TumorSim 1.2.2 on my Scientific Linux version 2.6.32 (x86-64 architecture), I got the same error as Flavio's:

./tumorsim_1.2.2_linux64: /usr/lib/x86_64-linux-gnu/libstdc++.so.6: version 'GLIBCXX_3.4.15' not found (required by ./tumorsim_1.2.2_linux64)
./tumorsim_1.2.2_linux64: /lib/x86_64-linux-gnu/libc.so.6: version 'GLIBC_2.14' not found (required by ./tumorsim_1.2.2_linux64) 

On Win7, I was able to run it after updating Visual Studio to Visual Studio 2012 (Express).  This solved the msvcp110.dll-error for me.

Regards,

Tom

Hi Tom,
thanks for your help.
release 1.2.2 works on kernel Linux Ubuntu 12.x

regards

Flavio

Hi Flavio and Tom,

I have uploaded a new version with statically linked libraries (tumorsim_1.2.2_static_linux64.zip) that should help with the library version conflicts on Linux.

Running the Windows version should not require installation VS 2012 express. You can try installing the VS 2012 redistributable package: http://www.microsoft.com/en-us/download/...

Thanks for the feedback,
Marcel