[Rtk-users] fdk artefact, sourceOffsetY

[Mathieu DUPONT]

Hi Simon, 

you are right, it works now. Thanks for your reactivity. 

----- Mail original -----

De: "Simon Rit" <simon.rit at creatis.insa-lyon.fr> 
À: "Mathieu DUPONT" <mdupont at cppm.in2p3.fr> 
Cc: rtk-users at public.kitware.com 
Envoyé: Lundi 9 Février 2015 15:08:12 
Objet: Re: [Rtk-users] fdk artefact, sourceOffsetY 

Hi Mathieu, 
Thanks for the observation. The truth is that we haven't tested thoroughly this kind of geometry because I generally prefer to keep the source in the central plane... Still, I have looked into the artifact that you pointed out, which is only visible with a narrow window / level, and found the cause which was a bug. I think this will fix it: 
https://github.com/SimonRit/RTK/commit/b0dd88821b751671d32ef0fa2f1ff2ae85a4083c 
Indeed, this only concerns FBP reconstruction, SART and other iterative algorithms should not be impacted. Let us know if you observe some other surprising artifacts. 
Thanks again, 
Simon 

On Fri, Feb 6, 2015 at 6:27 PM, Mathieu DUPONT < mdupont at cppm.in2p3.fr > wrote: 


Hello rtk users, 

I try to use RTK for reconstruction in my lab but I encounter some problems. 

I try to reconstruct a simulated spheroid. I join my phantom text file (SheppLogan.txt). 

If I set source_y=0, I get a correct (I guess) reconstruction, see the joined figure1.png. I use the followind set of commands : 

rtksimulatedgeometry -n 180 --sdd=508.1658133423730 --sid=242.7169928585076 --proj_iso_x=-53.500206362627239 --proj_iso_y=-10.768620138555379 --source_y=0 -o geometry.xml && rtkprojectgeometricphantom -g geometry.xml -o /tmp/tmpfs/projections.mha --spacing 1 --dimension 512 --phantomfile SheppLogan.txt && rtkfdk -v -p /tmp/tmpfs/ -r projections.mha -o /tmp/tmpfs/fdk_ccd.mha -g geometry.xml --spacing 0.07 --origin -20,14,-20 --dimension 600,600,600 --hardware="cuda" 

But If I set source_y=15, I obtain some strange artefacts as you can see on figure2.png. Do you know the origin of this pattern (the vertical one) ? Is it normal ? Or my geometry is completely wrong ? If I use rtksart, this artefact doesn't exist. 

Thanks in advance and thanks for this nice toolkit. 

regards, 
Mathieu Dupont 






_______________________________________________ 
Rtk-users mailing list 
Rtk-users at public.kitware.com 
http://public.kitware.com/mailman/listinfo/rtk-users 






-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/rtk-users/attachments/20150212/033f1dd0/attachment-0009.html>