search

LLNL / LEAP

comprehensive library of 3D transmission Computed Tomography (CT) algorithms with Python API and fully integrated with PyTorch
https://leapct.readthedocs.io
MIT License
74 stars 8 forks source link

Can we intinialize the CT image when do iterative reconstruction methods on sinogram? #68

Open GreameLee opened 2 days ago

GreameLee commented 2 days ago

When I try to add this process in diffusion model. I need insert LEAP with another reconstruction methods. So I need the iterative reconstruction method such as SART can take the processed CT images as the initial image to use its corresponding sinogram reconstruct images. So can I initialize the image?

kylechampley commented 2 days ago

Yes, you definitely can do this with LEAP. In LEAP, when you run any iterative reconstruction algorithm the first argument is the CT projection data (sinogram) and the second argument is the CT volume data. This volume data can have values from a different reconstruction method or it can be all zeros for a fresh start- whatever you want. There are several demo scripts that demonstrate this being done, one for example is this one: d29_filter_sequence.py.

I assume that since you are asking this question that you were able to install LEAP. Would you mind letting me know how you resolved your issue?

GreameLee commented 1 day ago

Yes, I have replied to that issue. And there is another question about the CT reconstruction algorithm. I see there is an iterative algorithm called OS-SART in the readme file but I can not find it in code. Is this method available to the public now?

kylechampley commented 1 day ago

In the code it is just called SART.