search

CERN / TIGRE

TIGRE: Tomographic Iterative GPU-based Reconstruction Toolbox
BSD 3-Clause "New" or "Revised" License
550 stars 182 forks source link

Bug in detector offset geometry reconstruction in FDK #382

Open yidu-bjcancer opened 2 years ago

yidu-bjcancer commented 2 years ago

Expected Behavior

Bug in the latest version of FDK for detector offset geometry reconstruction, only in certain geometry (maybe rotation direction? unclear yet)

Actual Behavior

It seems the detector offset weighting is incorrect.

Code to reproduce the problem (If applicable)

I think someone modified the FDK.m function, particularly the weighting session, and there is a bug there. When I replaced the latest FDK.m with an old version, the bug disappeared.

Specifications

Report first. I can fix this myself when I have time.

AnderBiguri commented 2 years ago

There has been ineed some changes in the weigting, but tests showed better recon with the changes. If you have troubles fixing it, can you post your geometry, so I can compare?

yidu-bjcancer commented 2 years ago

There has been ineed some changes in the weigting, but tests showed better recon with the changes. If you have troubles fixing it, can you post your geometry, so I can compare?

This is the geometry to trigger the bug. geo.zip

shaohua-zhi commented 1 year ago

@AnderBiguri @yidu-bjcancer Hi, I also met a problem with the same Varian half-fan geometry. after adding wang pre-weighting(not sure which version), the reconstructed image has a circle dark drop at the center. could you help me fix it? image

AnderBiguri commented 1 year ago

Hi @shaohua-zhi , thanks for the message.

Can you verify which exact version of TIGRE are you using? In the latest release (2.4) a big like this existed, but in the current latests commit in master it does not.

What you see may be a different bug, but I want to first verify that, and if it is, I would need to have a way to reproduce it, so I can fix it. Can you share (privately if you want) the data, if you are using the latest version of TIGRE?

AnderBiguri commented 1 year ago

@shaohua-zhi I have just released TIGRE 2.5, check if the error is still there.

shaohua-zhi commented 1 year ago

@AnderBiguri Thanks a lot! It works now. may I ask if are there any utilities in TIGRE for ElektaCBCT data?

yidu-bjcancer commented 1 year ago

@shaohua-zhi Sorry for this late reply. I guess the bug was in the Python version FDK, since the Matlab version works well for me.

As for TIGRE for Elekta, I am sorry that clinical work piles up and I can not spare enough time to write an ElektaCBCT module (probably until the end of this year). As an alternative, you may use RTK, which provides a specific module for ElektaCBCT (however I don't have any user experience with it).

AnderBiguri commented 1 year ago

@shaohua-zhi I have loaded from an Elekta machine in the past, but it was just dicom files. Would be happy to help, but I don't have data from such machine, so some data would need to be shared with me.