mri-synthesis / LG-Net

Official repository for MICCAI2021-LG_Net

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

LG-Net

Official PyTorch implementation of LG-Net: Lesion Gate Network for Multiple Sclerosis Lesion Inpainting, MICCAI 2021

Paper: LG-Net

LG-Net

Dependencies

  • Python >= 3.6

Pytorch version

  • torch >= 1.5.0
  • torchvision >= 0.6.0

Data

Please refer to the official website (or project repo) for license and terms of usage. Some preprocessing have been applied to the original data before feeding into our data loader. Please find the details in our paper.

IXI

Train Please refer to the training script in the scripts folder

python3 ../train.py \
--dataroot [path_to_dataset] \
--checkpoints_dir [path_to_saved_weights] \
--gpu_ids 0 \
--name [saved_name] \
--model lesion_inpaint_lgc \
--input_nc 6 \
--output_nc 3 \
--init_type kaiming \
--dataset_mode brain \
--num_threads 8 \
--batch_size 24 \
--beta1 0.99 \
--lr 0.0001 \
--lambda_lgc 0.1 \
--lambda_lesion 10 \
--lambda_tissue 1 \
--n_epochs 500 \
--print_freq 1000 \
--save_latest_freq 5000 \
--save_epoch_freq 100 \

Test Please refer to the testing script in the scripts folder

python3 ../test.py \
--dataroot [path_to_dataset] \
--checkpoints_dir [path_to_saved_weights] \
--gpu_ids 0 \
--name ixi_gate_lgc \
--model lesion_inpaint_lgc \
--input_nc 6 \
--output_nc 3 \
--pad_to_size -1 \
--view ax

About

Official repository for MICCAI2021-LG_Net


Languages

Language:Python 99.2%Language:Shell 0.8%