Possible problem with module axial spacing modelization in the recon utility


I think I have found where was the problem that created the error I talked about in my previous mail.

The line 644-645 of iScannerPET.cc are:
FLTNBLUT size_trs_rsector = nb_trs_mod*size_trs_mod + (nb_trs_submod-1)*gap_trs_mod;
FLTNBLUT size_axl_rsector = nb_axl_mod*size_axl_mod + (nb_axl_submod-1)*gap_axl_mod;
But it would make more sense if they were:
FLTNBLUT size_trs_rsector = nb_trs_mod*size_trs_mod + (nb_trs_mod-1)*gap_trs_mod;
FLTNBLUT size_axl_rsector = nb_axl_mod*size_axl_mod + (nb_axl_mod-1)*gap_axl_mod;

I recompiled Castor with that modification and the problem I observed previously are no longer there. I didn't make any further tests. However, I think it is pretty straightforward, thanks to the name of the variables.

By the way, is castor hosted somewhere like github and, in the affirmative, should I make a pull request?

Have a nice day,
Maxime Toussaint

Hello Maxime,

Many thanks for your feedback and corrections. The wrong variables were indeed used in these lines. They have been corrected and the current version on the download page (http://www.castor-project.org/download_castor) has been updated.

Regarding the geometry, there could also be miscalculations of the gap dimensions using the conversion tools, due to the way they are calculated from the GATE mac files (this could happen if the dimensions of the rsectors as set in the mac files don't match the actual dimensions of all the elements (modules/crystals and their relative gaps) composing them). This does not seem to be the case with your data though. Anyway, this should be addressed in the next release of CASToR which will come soon.

There is no public repository for the moment for Castor, but this question has been raised before and we will have a look on it after the next release.

Thanks again, and kind regards