Sparse2Noise

Low-dose synchrotron X-ray tomography without high-quality reference data

Paper: to be published

Setup Environemnets:

create your virtual environment and install dependencies:

1. Open a terminal or command prompt.
2. Navigate to the directory where you want to create the virtual environment (usually the path contains your script).
3. Run the following command to create a new virtual environment: 
   /usr/bin/python3.6 -m venv venv
4. Activate the virtual environment:
   source /.../venv/bin/activate
6. Once the virtual environment is activated, run the following command to install the dependencies from the provided requirements.txt file:
   pip install --upgrade pip
   pip install -r requirements.txt
        or pip install numpy tensorflow imageio scipy tifffile

Dataset (h5):

Test data 241train-scaffolds-sparse2noise.h5 can be downloaded from here:

h5dump -H 241train-scaffolds-sparse2noise.h5
HDF5 "241train-scaffolds-sparse2noise.h5" {
GROUP "/" {
   DATASET "test_gt" {
      DATATYPE  H5T_STD_U8LE
      DATASPACE  SIMPLE { ( 5, 256, 256 ) / ( 5, 256, 256 ) }
   }
   DATASET "test_ns" {
      DATATYPE  H5T_STD_U8LE
      DATASPACE  SIMPLE { ( 5, 256, 256 ) / ( 5, 256, 256 ) }
   }
   DATASET "train_gt" {
      DATATYPE  H5T_STD_U8LE
      DATASPACE  SIMPLE { ( 241, 256, 256 ) / ( 241, 256, 256 ) }
   }
   DATASET "train_ns" {
      DATATYPE  H5T_STD_U8LE
      DATASPACE  SIMPLE { ( 241, 256, 256 ) / ( 241, 256, 256 ) }
   }
}
}

train_ns: Normal-flux sparse-view CT reconstruction images for training

train_gt: Low-flux full-view CT reconstruction images for training

test_ns: Normal-flux sparse-view CT reconstruction images for test

test_gt: Low-flux full-view CT reconstruction images for test

Train:

run main.py, an example:
```
python main.py -h5fn 241train-scaffolds-sparse2noise.h5
```
Now you should have a folder /Output/<your dataset's name>

Example:

Reference