Hello Everyone,
I have tried to reconstruct my images using OSL. I have tried various commands to do so, however, I am not able to get it to work. For example, I typed the following in the Terminal:
castor-recon -df somedata.cdh -opti OSL,1,1e-10,0.01,-1 -pnlt MRP:/path/to/config/optimizer/MRP.conf -it 10:4 -proj joseph -dim 100,100,100 -vox 1,1,1 -dout sth
However I got the error that reads as follows: “A problem occurred while checking penalty parameters”.
I modified the MRP.conf such that only 6-nearest is uncommented and everything else is commented. (I tried many other ways as well)
I would really appreciate it if someone could help me on how to pass these parameters for OSL and guide me to figure out my mistake.
Also another quick question: If I have two point spread functions, PSF1 and PSF2 (one for positron range for example) how can I pass them to castor? Is correct to pass as follows?
-conv PSF1,X1,X2,X3::when -conv PSF2, Y1,Y2,Y3::when
Best,
Seyyed
I have found the answer on how to reconstruct using OSL with penalty terms. I needed to specify the strength of the penalty by setting -pnlt-beta. Here is the final command:
castor-recon -df somedata.cdh -opti OSL,1,1e-10,0.01,-1 -pnlt MRP:/path/to/config/optimizer/MRP.conf -pnlt-beta 1 -it 10:4 -proj joseph -dim 100,100,100 -vox 1,1,1 -dout sth
I found this looking at “castor-recon -help-algo”.
Best regards,
Seyyed
1 Like
Dear Seyyed,
Is there any difference between the image reconstructed using OSL and the image reconstructed using MLEM? I reconstructed my own data obtained from self-designed PET scanner using MLEM and OSL with different prior model, but I cannot see any difference in the image. I’d like to know if you’ve encountered a similar situation.
Best regards,
Zhijun Zhao
Dear Zhijun,
I think it depends on the parameters that you choose. For me, there was not a huge difference but I observed some differences, in particular, the edge artifacts and deconvolution. Please let me know if there is anything that I can help. However, I am yet new to CASToR.
Best regards,
Seyyed