MoG-QSM: Model-based Generative Adversarial Deep Learning Network for Quantitative Susceptibility Mapping.
MoG-QSM was proposed by Ruimin Feng and Dr. Hongjiang Wei. It reconstructs high quality susceptibility maps from tissue phase.
###Environmental Requirements:
- Python 3.6
- Tensorflow 1.15.0
- Keras 2.2.5
###Files descriptions
MoG-QSM contains the following folders:
-
data: It provides three types of test data: healthy data from Siemens Prisma scanner, multiple sclerosis data, and 2016 QSM Challenge data.
-
logs/last.h5: A file that contains the weights of the trained model
-
model/MoG_QSM.py : This file contains the functions to create the Model-based Generative Adversarial Network proposed in our paper
-
test: It contains test_tools.py and test_demo.py.
test_tools.py offers some supporting functions for network testing such as image patch stitching, dipole kernel generation, etc. test_demo.py shows how to perform network testing with data from the "data" folder -
train: It contains train_gen.py, train_joint.py and utils.py.
train_gen.py: This is the code for generator training
train_joint.py: This is the code for generator and discriminator jointly training
utils.py: It offers some supporting functions for network training -
NormFactor.mat: The mean and standard deviation of our training dataset for input normalization.
###Usage
##Test
- You can run test_demo.py directly to test the network performance on the provided data. The results will be in the same directory as the input data
- For test on your own data. You can use "model_test" function as shown in test_demo.py files
##train
- If you want to train MoG-QSM by yourself. train_gen.py and train_joint.py can be used as a reference.