Epi_correction

This repository is a torch implementation of "Unsupervised cycle-consistent network using restricted subspace field map for removing susceptibility artifacts in EPI"

Pretrained Models

you can download a pretrained version of FlowNetS (from caffe, not from pytorch) here. This folder also contains trained networks from scratch. Thanks to Kaixhin.

Note on networks loading

Directly feed the downloaded Network to the script, you don't need to uncompress it even if your desktop environment tells you so.

Prerequisite

these modules can be installed with pip

pytorch >= 1.2
tensorboard-pytorch
tensorboardX >= 1.4
spatial-correlation-sampler>=0.2.1
imageio
argparse
path.py

pip install -r requirements.txt

Default HyperParameters provided in main.py are the same as in the caffe training scripts.

Example usage for UCRSF-Net :

python main.py

Visualizing training

Tensorboard-pytorch is used for logging. To visualize result, simply type

tensorboard --logdir=/path/to/checkoints

Running inference on a set of image pairs

If you need to run the network on your images, you can download a pretrained network here and launch the inference script on your folder of image pairs.

Your folder needs to have all the images pairs in the same location, with the name pattern

{image_name}1.{ext}
{image_name}2.{ext}

python3 run_inference.py /path/to/images/folder /path/to/pretrained

Acknowledge

Thanks to the strong and flexible FlowNetPytorch codebase maintained by Dosovitskiy et al.

yinggo / Epi_correction