ginobilinie / medSynthesisV1

This is a copy of package for medical image synthesis work with LRes-ResUnet and GAN (wgan-gp) in pytorch framework
MIT License
183 stars 45 forks source link

lambda parameter settings #23

Closed 921974496 closed 4 years ago

921974496 commented 4 years ago

lambda_AD lambda_gdl lossBase lambda_D_WGAN_GP How to set these parameters? Is it by setting these parameters that the loss is in the same order of magnitude?

ginobilinie commented 4 years ago

Honestly speaking, it's hard to answer. You can try several values (like what we give the the code). But I donot think we should set the loss to make them the same order of magnitude.

921974496 commented 4 years ago

Questions about ACM

“ The only difference is about the input to the generator, which concatenates the source MRI patch and the synthetic target patch since the 1st iteration of ACM.“

I didn't find “concatenates the source MRI patch and the synthetic target patch” in the code

921974496 commented 4 years ago

I want to ask about what software is used to implement difference map ?

For example: difference map between the synthetic CT and ground truth CT

ginobilinie commented 4 years ago

Do you mean the visualization of difference map? I use matlab and ITKSnap.

921974496 commented 4 years ago

Do you mean the visualization of difference map? I use matlab and ITKSnap.

Thank you very mach ,Can you send me the MATLAB code? If it's not convenient, don't send it

921974496 commented 4 years ago

Do you mean the visualization of difference map? I use matlab and ITKSnap.

And how is ACM implemented?