help > def2.mha and def2_inverse.mha incorrect?
Showing 1-1 of 1 posts
Results per page:
Mar 3, 2014  02:03 PM | Tom Haeck
def2.mha and def2_inverse.mha incorrect?
Hi all,

- I seem to have a problem interpreting the meaning of *_def2.mha and *_def2_inverse.mha as output of the TumorSim simulator.  I assumed that both mappings were the inverse mapping from one another.  This would mean that if you concatenate them, regardless of the order in which you concatenate them, the result should always be the identity transform (small deviations from the identity transform would be possible because of numerical instability for calculating the inverse or interpolation errors in the concatenation).  However, this is never the case for the tumors that I've simulated.  

The deviation is quite large and causes weird phenomena:  *def2.mha is used to resample the tissue class probabilities into the space of the tumor image.  This is reflected into the file *_warped_labels2.mha  However, if you compare the location of the tumor in the files *_warped_labels2.mha and *_T1.mha, the location is always quite different and even seldom overlapping.  I assume this is caused by the wrong calculation of *_def2.mha from *_def2_inverse.mha?  Is someone else experiencing this problem or does it occur only with a certain choice of parameters?

- I'm still not able to manage to run the TumorSim software on our Linux-cluster.  The GLIBC-libraries are a little outdated and the department is not planning to update them because it is quite costly.  It is also not possible for me to run it locally on my MacOSX operating system.  Would it be somehow possible to make the software open source?  This way I can compile it on my own platform?  It would be a great great great help:-)