initial commit

.gitignore

@@ -0,0 +1,113 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+env/
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+.hypothesis/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# pyenv
+.python-version
+
+# celery beat schedule file
+celerybeat-schedule
+
+# SageMath parsed files
+*.sage.py
+
+# dotenv
+.env
+
+# virtualenv
+.venv
+venv/
+ENV/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+
+#checkpoints and tensorboards
+events*
+checkpoint*
+graph*
+
+#data
+raw*
+tmp*
+.zip
+*.zip
+data

README.md

@@ -1 +1,46 @@
-# unsupGAN-release
+# unsupGAN-release
+
+# Unsupervised MRI Reconstruction
+
+Unsupervised MRI Reconstruction with Generative Adversarial Networks
+
+- 2020 Elizabeth Cole, Stanford University ([email protected])
+
+## Setup
+
+Make sure the python requirements are installed
+
+    pip3 install -r requirements.txt
+
+The setup assumes that the latest Berkeley Advanced Reconstruction Toolbox is installed [1]. The scripts have all been tested with v0.4.01.
+
+## Data preparation
+
+We will first download data, generate sampling masks, and generate TFRecords for training. The datasets downloaded are fully sampled volumetric knee scans from mridata [2]. The setup script uses the BART binary. In a new folder, run the follwing script:
+
+    python3 mri_util/setup_mri.py -v
+
+## Training/Testing Unsupervised GAN
+
+The training of the unsupervised GAN can be ran using the following script:
+    python3 train_unsupervised.py dataset_dir model_dir
+where dataset_dir is the folder where the knee datasets were saved to,
+and model_dir will be the top directory where the models will be saved to.
+
+Testing can be ran using:
+    python3 test_unsupervised.py dataset_dir model_dir
+
+## Training/Testing Supervised GAN
+
+The training of the supervised GAN can be ran using the following script:
+    python3 train_supervised.py dataset_dir model_dir
+where dataset_dir is the folder where the knee datasets were saved to,
+and model_dir will be the top directory where the models will be saved to.
+
+Testing can be ran using:
+    python3 test_supervised.py dataset_dir model_dir
+
+## Questions/Issues
+
+For any issues or questions, please open an issue on the github repo or contact
+Elizabeth at [email protected].

create_masks.py

@@ -0,0 +1,14 @@
+from mri_util import mri_prep
+
+dir_out = "/home_local/ekcole/knee_masks"
+mri_prep.create_masks(
+    dir_out,
+    shape_y=320,
+    shape_z=256,
+    verbose=True,
+    acc_y=(1, 2, 3, 4),
+    acc_z=(1, 2, 3, 4),
+    shape_calib=1,
+    variable_density=True,
+    num_repeat=4,
+)

masks_train_acc.py

@@ -0,0 +1,34 @@
+from mri_util import mri_prep
+acc_y = 2
+acc_z = 7
+total_acc = acc_y*acc_z
+dir_out = "/home_local/ekcole/knee_masks_%d" % total_acc
+print(dir_out)
+mri_prep.create_masks(
+    dir_out,
+    shape_y=320,
+    shape_z=256,
+    verbose=True,
+    acc_y=[acc_y],
+    acc_z=[acc_z],
+    shape_calib=1,
+    variable_density=True,
+    num_repeat=3,
+)
+
+acc_y = 7
+acc_z = 2
+total_acc = acc_y*acc_z
+dir_out = "/home_local/ekcole/knee_masks_%d" % total_acc
+print(dir_out)
+mri_prep.create_masks(
+    dir_out,
+    shape_y=320,
+    shape_z=256,
+    verbose=True,
+    acc_y=[acc_y],
+    acc_z=[acc_z],
+    shape_calib=1,
+    variable_density=True,
+    num_repeat=3,
+)

mri_util/cfl.py

@@ -0,0 +1,66 @@
+# Copyright 2013-2015. The Regents of the University of California.
+# All rights reserved. Use of this source code is governed by
+# a BSD-style license which can be found in the LICENSE file.
+#
+# Authors:
+# 2013 Martin Uecker <[email protected]>
+# 2015 Jonathan Tamir <[email protected]>
+
+import numpy as np
+
+
+def read_hdr(name, order="C"):
+    # get dims from .hdr
+    h = open(name + ".hdr", "r")
+    h.readline()  # skip
+    l = h.readline()
+    h.close()
+    dims = [int(i) for i in l.split()]
+    if order == "C":
+        dims.reverse()
+    return dims
+
+
+def read(name, order="C"):
+
+    dims = read_hdr(name, order)
+
+    # remove singleton dimensions from the end
+    n = np.prod(dims)
+    dims_prod = np.cumprod(dims)
+    dims = dims[: np.searchsorted(dims_prod, n) + 1]
+
+    # load data and reshape into dims
+    d = open(name + ".cfl", "r")
+    a = np.fromfile(d, dtype=np.complex64, count=n)
+    d.close()
+    return a.reshape(dims, order=order)  # column-major
+
+
+def readcfl(name):
+    return read(name, order="F")
+
+
+def write(name, array, order="C"):
+    h = open(name + ".hdr", "w")
+    h.write("# Dimensions\n")
+    if order == "C":
+        for i in array.shape[::-1]:
+            h.write("%d " % i)
+    else:
+        for i in array.shape:
+            h.write("%d " % i)
+    h.write("\n")
+    h.close()
+
+    d = open(name + ".cfl", "w")
+    if order == "C":
+        array.astype(np.complex64).tofile(d)
+    else:
+        # tranpose for column-major order
+        array.T.astype(np.complex64).tofile(d)
+    d.close()
+
+
+def writecfl(name, array):
+    write(name, array, order="F")