Skip to content

tbuikr/fastMRI

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

28 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Run training ESSIMTV + Aug + Dilation + Resnet +128

python train_unet_essimtv_aug_scSE.py --challenge singlecoil --data-path /media/toanhoi/88f64337-6c11-4924-a165-ca6fefb38002/home/toanhoi/KneeData --exp-dir checkpoint --netG unet_upsampling_dilation --batch-size 16 --num-chans 128

Run evaluate ESSIMTV + Aug + Dilation + Resnet +128 without TTA

python run_unet_transpose.py --data-path ../../Knee_fastMRI --data-split val --checkpoint ./checkpoint/model.pt --challenge singlecoil --out-dir /media/toanhoi/Data/Knee_fastMRI/reconstructions_val --mask-kspace --batch-size 16 --netG unet_upsampling_dilation 
 python evaluate.py --target-path ../../Knee_fastMRI/singlecoil_val --predictions-path /media/toanhoi/Data/Knee_fastMRI/reconstructions_val --challenge singlecoil

Run evaluate ESSIMTV + Aug + Dilation + Resnet +128 with TTA

python run_unet_transpose.py --data-path ../../Knee_fastMRI --data-split val --checkpoint ./checkpoint/model.pt --challenge singlecoil --out-dir /media/toanhoi/Data/Knee_fastMRI/reconstructions_val --mask-kspace --batch-size 16 --netG unet_upsampling_dilation  --tta 1
 python evaluate.py --target-path ../../Knee_fastMRI/singlecoil_val --predictions-path /media/toanhoi/Data/Knee_fastMRI/reconstructions_val --challenge singlecoil

-----------------------------------Old --------------------------------

Run ESSIMTV + AUG + Attention

python train_unet_essimtv_aug_scSE.py --challenge singlecoil --data-path ../../Knee_fastMRI/ --exp-dir checkpoint --netG unet_upsampling_scSE --batch-size 16 --num-chans 128

Run L1+SSIm+TV+newnet+res+aug

CUDA_VISIBLE_DEVICES=0 python train_unet_l1cssimtv_unet_transpose.py --challenge singlecoil --data-path ../../Knee_fastMRI/ --exp-dir checkpoint --aug True --batch_size 16 --netG unet_transpose_res

Run L1+ssim+tv+newunet+aug

CUDA_VISIBLE_DEVICES=0 python train_unet_l1cssimtv_unet_transpose.py --challenge singlecoil --data-path ../../Knee_fastMRI/ --exp-dir checkpoint --aug True --batch_size 16

Run L1+ssim+tv+newunet

python train_unet_l1cssimtv_unet_transpose.py --challenge singlecoil --data-path ../../Knee_fastMRI/ --exp-dir checkpoint

Run evaluation on new network

python run_unet_transpose.py --data-path ../../Knee_fastMRI --data-split val --checkpoint ./checkpoint/best_model.pt --challenge singlecoil --out-dir /media/toanhoi/Data/Knee_fastMRI/reconstructions_val --mask-kspace --batch-size 16 --netG unet_transpose
python evaluate.py --target-path ../../Knee_fastMRI/singlecoil_val --predictions-path /media/toanhoi/Data/Knee_fastMRI/reconstructions_val --challenge singlecoil

Run test on new network

python run_unet_transpose.py --data-path ../../Knee_fastMRI/ --data-split test --checkpoint ./checkpoint/best_model.pt --challenge singlecoil --out-dir /media/toanhoi/Data/Knee_fastMRI/reconstructions_test --batch-size 16 --netG unet_transpose

Run with modified unet using up-sampling

python train_unet_l1cssimtv_unet_transpose.py --challenge singlecoil --data-path ../../Knee_fastMRI/ --exp-dir checkpoint --netG unet_upsampling

=============BASELINE=========

Run train l1+ssim

python train_unet_l1cssim.py --challenge singlecoil --data-path ../../Knee_fastMRI/ --exp-dir checkpoint

RUn train l1+ssim+cGAN +resume G

python train_unet_l1cssim_dis.py --challenge singlecoil --data-path ../../Knee_fastMRI/ --exp-dir checkpoint --resume --checkpoint ./checkpoint/best_model.pt

RUn train l1+ssim+cGAN +noresume G

python train_unet_l1cssim_dis.py --challenge singlecoil --data-path ../../Knee_fastMRI/ --exp-dir checkpoint

Run test

python run_unet.py --data-path ../../Knee_fastMRI/ --data-split test --checkpoint ./checkpoint/best_model.pt --challenge singlecoil --out-dir /media/toanhoi/Data/Knee_fastMRI/reconstructions_test

Run val

python run_unet.py --data-path ../../Knee_fastMRI --data-split val --checkpoint ./checkpoint/best_model.pt --challenge singlecoil --out-dir /media/toanhoi/Data/Knee_fastMRI/reconstructions_val --mask-kspace
python evaluate.py --target-path ../../Knee_fastMRI/singlecoil_val --predictions-path /media/toanhoi/Data/Knee_fastMRI/reconstructions_val --challenge singlecoil

===================Val baseline=======================

Baseline

MSE = 1.608e-10 +/- 3.858e-10 NMSE = 0.04453 +/- 0.05602 PSNR = 30.42 +/- 5.805 SSIM = 0.6755 +/- 0.2835

Baseline+ssim

MSE = 1.506e-10 +/- 3.54e-10 NMSE = 0.04342 +/- 0.05655 PSNR = 30.6 +/- 5.964 SSIM = 0.6901 +/- 0.2718

About

No description, website, or topics provided.

Resources

License

Code of conduct

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published