hyn2028/tpdm

Official code for "Improving 3D Imaging with Pre-Trained Perpendicular 2D Diffusion Models" (TPDM)

tpdm's Issues

Training Model Tips

Hello!
I'm trying to use tpdm for generating a 3D Unconditional Volume using two diffusion models trained with knee volumes.
One of the models was trained with coronal slices and the other with axial slices. When I use those checkpoints, after 2000 sampling steps, the images (Volume) generated are completely noisy, almost white noise.
Do you have any tips on what is going wrong? Is there a specific orientation of the slices used to train the models?
My dataset was also small (10 volumes). Could that be one of the reasons for the issues? However, when I use the LDCT checkpoints, the TPDM is able to generate a 3D unconditional volume, and those models were also trained with a small dataset

Best regards.

Super Resolution Factor

Hello. I was checking the code and noticed that the factor for MR-ZSR is only defined from 2 to 5. Was that an intentional choice or is the code also prepared for greater factors. How can the matrices M, associated with the factor, be computed? Thanks and best regards.

TOO slow inference speed

I am testing the code run_tpdm_svct.py, it needs almost 15 hours for a single case (batchsize=8 A10080G). It's too slow, any methods to accelerate?

Request for AAPM CT checkpoint in the paper.

Hello, your paper showcases excellent reconstruction performance of sparse view CT, which has piqued my interest. But what is provided in the code is a checkpoint for another set of simulated data, not for the AAPM CT dataset. I'm not sure if it's convenient to provide the pretrained checkpoint for AAPM CT. I also noticed that your previous work on DiffuseMBIR provided checkpoints for AAPM, and if TPDM could be provided, it would be more helpful in forming a comparison. Thank you.

Is there any strategy to set the dps_weight in MRI reconstruction

Score-SDE Dependecies

Hello.
Do you happen to know the exact dependencies versions of the score_sde code you used for training? There seems to be an error when installing the dependencies using the original repository.
Thanks and best regards.

TRAIN

train

I'm implementing your code.Is it convenient for you to provide your CT training and test data set files? If it is ok.This my email:[email protected] would appreciate it