SegSRGAN 2.1.1

Segmentation and super resolution GAN network

SegSRGAN

This algorithm is based on the method proposed by Chi-Hieu Pham in 2019. More information about the SEGSRGAN algorithm can be found in the article

Installation

User (recommended)

The library can be installed using Pypi

pip install SegSRGAN

NOTE: We recommend to use virtualenv

Developer

First, clone the repository. Use the make to run the testsuite or yet create the pypi package.

git clone [e-mail]:koopa31/SegSRGAN.git

make test
make pkg

make

Perform a training

python SegSRGAN_train_avec_base_test.py --newlowres --csv --snapshot_folder --dice_file --mse_file --folder_training_data

• csv file which contains the paths to the files for the training. They are divided into two categories: train and test. As a consequence, it must contain 3 columns respectively called: HR_image, Label_image and Base (which can be equal to either Train or Test).
• dice_file CSV file where to store the DICE at each epoch (string)
• mse_file MSE file where to store the DICE at each epoch (string)
• folder_training_data folder which contains the training images database (string)
• epoch number of training epochs (integer)
• batchsize number of patches per mini batch (integer)
• snapshot_folder how often the weights are saved on the disk (for instance if equal to 2, the weights are saved on the disk one epoch in two)(integer)}
• numcritic how many times we train the discriminator before training the generator (integer)

But it is also possible to continue a training from its saved weight, adding the following parameters:

• initepoch number of the epoch from which the training will continue (integer)
• weights path to the saved weights from which the training will continue (string)

Two very important parameters to set the structure of the network are:

• kernelgen Number of output channel of the first convolutional layer of the generator (see \hyperref[architecture]{ section \ref*{architecture}})
• kerneldis Number of output channel of the first convolutional layer of the discriminator

Perform a segmentation

from SegSRGAN.SegSRGAN.Function_for_application_test_python3 import segmentation

segmentation(input_file_path, step, NewResolution, path_output_cortex, path_output_HR, weights_path, patch=None, spline_order=3, by_batch=False, is_conditional=False)

Where:

• input_file_path is the path of the image to be super resolved and segmented
• step is the shifting step for the patches
• NewResolution is the new z-resolution we want for the output image
• path_output_cortex output path of the segmented cortex
• path_output_HR output path of the super resolution output image
• weights_path is the path of the file which contains the pre-trained weights for the neural network
• patch is the size of the patches
• spline_order for the interpolation
• by_batch is to enable the by-batch processing

Segmentation of a set of images with several step and patch values

In order to facilitate the segmentation of several images, you can run SegSRGAN/SegSRGAN/job_model.py:

python job_model.py --path --patch --step --result_folder_name --weights_path

The list of the paths of the images to be processed must be stored in a csv file.

Where:

• path Path of the csv file
• patch list of patch sizes
• step list of steps
• result_folder_name Name of the folder containing the results

Example of syntax for step and patch setting:

--patch 64,128

--step 32 64,64 128

In this example we run steps 32 and 64 for patch 64 and steps 64 and 128 for patch 128. Be careful to respect the exact same spaces.