DAGAN

This is the implementation code for "DAGAN: Deep De-Aliasing Generative Adversarial Networks for Fast Compressed Sensing MRI Reconstruction". Paper Link

The code was written by Simiao Yu and Hao Dong.

If you use this code for your research, please cite our paper.

If you have any questions, please contact Simiao Yu (simiao.yu13@imperial.ac.uk) or Guang Yang (g.yang@imperial.ac.uk).

Prerequisites

The original code is in python 3.5 under the following dependencies: 1. tensorflow (v1.1.0) 2. tensorlayer (v1.7.2) 3. easydict (v1.6) 4. nibabel (v2.1.0) 5. scikit-image (v0.12.3)

Code tested in Ubuntu 16.04 with Nvidia GPU + CUDA CuDNN.

How to use the code

1. Prepare data

   1) Data used in this work are publicly available from the MICCAI 2013 grand challenge (link). We refer users to register with the grand challenge organisers to be able to download the data. 2) Download training and test data respectively into data/MICCAI13_SegChallenge/Training_100 and data/MICCAI13_SegChallenge/Testing_100 (We randomly included 100 T1-weighted MRI datasets for training and 50 datasets for testing) 3) run 'python data_loader.py' 4) after running the code, training/validation/testing data should be saved to 'data/MICCAI13_SegChallenge/' in pickle format.

2. Download pretrained VGG16 model

   1) Download 'vgg16_weights.npz' from this link 2) Save 'vgg16_weights.npz' into 'trained_model/VGG16'

3. Train model 1) run 'CUDA_VISIBLE_DEVICES=0 python train.py --model MODEL --mask MASK --maskperc MASKPERC' where you should specify MODEL, MASK, MASKPERC respectively:

• MODEL: choose from 'unet' or 'unet_refine'

• MASK: choose from 'gaussian1d', 'gaussian2d', 'poisson2d'

• MASKPERC: choose from '10', '20', '30', '40', '50' (percentage of mask)

4. Test trained model

   1) (to be finished soon)

Results

Please refer to the paper for detailed results.