CMLcGAN

This is the code in our paper 'The diagnosis of chronic myeloid leukaemia with deep adversarial learning'

System Requirements

We highly recommend using an nVIDIA GPU with >=8GB for testing forward of CML segmentation.

Installation

Requirments

• Python3.6+ (The installation tutorial can be found on page: https://www.python.org/getit/)
• matplotlib
• numpy
• sklearn
• skimage (pip install scikit-image)
• Pytorch1.0+ and torchversion (The installation tutorial can be found on page: https://pytorch.org/)

Usage

git clone https://github.com/zjuzzl/CML.git
cd CML

After all the requirements libraries are installed, let's get start. We recommend using Spyder as the code IDE. Just type 'pip install spyder' to install it.

Step1. CML segmentation

First of all, you need to put the pathological images to be segmented into a folder and create a folder to save the results. (example: 'data/img/', 'data/pred/') python test.py image_path result_save_path (For example: python test.py data/img/ data/pred/ ) You can find the segmented CML result image in 'result_save_path' path

Step2. Feature extraction

Modify the parameter in 'feature_extract.py', then run it, extracted features will be saved as a '.npy' file.

Step3. Feature selection and boxplot

feature_checking.py

Step4. Clinical validation and cross validation

clinical_analysis.py