yunboer/AR-classifier-and-AR-CycleGAN

AR-classifier-and-AR-CycleGAN

Introduction

This repository contains PyTorch code for the paper: Artifact Detection and Restoration in Histology Images with Stain-Style and Structural Preservation

This repository contains two new models: AR-classifier and AR-CycleGAN. The AR-classifier is designed to classify artifacts, while the AR-CycleGAN is capable of artifact recovery.

The AR-classifier is designed based on the ResNet18 architecture and is capable of classifying patches from whole slide images (WSIs) into three categories: "artifact," "normal," and "unrestorable."

The AR-CycleGAN is designed based on the CycleGAN and can remove artifacts from patches belonging to the "artifact" category.

Datasets

We provide two download links here for the our artifacts classification and CycleGAN dataset.

OneDrive

Baidu Cloud

Getting Started

Ensure your Python >= 3.8 (recommended 3.8)

1. Clone the repo

   git clone https://github.com/yunboer/AR-classifier-and-AR-CycleGAN.git
   

2. pip all the requirement packages.

   # if you want to try AR-classifier
   cd AR-classifier
   
   # if you want to try AR-CycleGAN
   cd AR-CycleGAN
   
   # pip install all the packages
   pip install -r requirements.txt
   

3. Prepare the yaml file for the RandStainNA

This document is currently being updated...