Refined RadImagenet - Conversion Tools

The RadImageNet dataset are available by request at https://www.radimagenet.com/)

This code designed to process the RadImagenet and convert to refined and stratified organization.

You can find the preprint paper here: Policy Gradient-Driven Noise Mask

If you use this code in your research, please cite our paper:

@article{yavuz2024policy,
  title={Policy Gradient-Driven Noise Mask},
  author={Yavuz, Mehmet Can and Yang, Yang},
  year={2024},
  eprint={2406.14568},
  archivePrefix={arXiv},
  primaryClass={eess.IV}
}

Performance Comparison of ResNet Models

This table compares the performance of ResNet models pretrained on 2D RadImagenet using regular and Two2Three convolution techniques across various metrics.

Features	Precision (macro)	Recall (macro)	F1 (macro)	Accuracy (balanced)	Accuracy (Average)
Resnet10t	0.4720	0.3848	0.3998	0.3848	0.7981
Resnet18	0.5150	0.4383	0.4545	0.4383	0.8177
Resnet50	0.5563	0.4934	0.5097	0.4934	0.8352

We highly recommend you to adap the code for benchmarking for other models:

https://github.com/pytorch/vision/tree/main/references/classification

ResNet Models and Weights

The model weights shared through https://huggingface.co/ogrenenmakine/RadImagenet

timm.create_model('resnet10t', num_classes=165)

The trained model are timm implementations.

Folder Structure

correction_masks/
data/
weights/
output/
source/
    correction_masks.tar.gz
    radimagenet.tar.gz
    RadiologyAI_test.csv
    RadiologyAI_train.csv
    RadiologyAI_val.csv
process.py
measure_acc_metrics.py

Files & Directories

• correction_masks/: Contains correction masks for the images.
• data/: Contains the extracted radiology images.
• weights/: Model weights containing folders.
• output/: Directory for output files.
• source/: Contains source files and datasets.
  • correction_masks.tar.gz: the file contains correction masks.
  • radimagenet.tar.gz: the original compressed RadImagenet file.
  • RadiologyAI_test.csv: CSV file for test dataset.
  • RadiologyAI_train.csv: CSV file for training dataset.
  • RadiologyAI_val.csv: CSV file for validation dataset.

Files

• process.py: Main script to process and organize the RadImagenet files.
• measure_acc_metrics.py: The script to measure accuracy metrics.

To create a GitHub README file with the instructions for using Git to clone the Hugging Face repository ogrenenmakine/Refined-RadImagenet, you can format it as follows:

Download Processing Files

This repository contains files from the Hugging Face repository ogrenenmakine/Refined-RadImagenet. Follow the instructions below to clone the repository using Git.

Prerequisites

If you haven't installed Git LFS yet, you can do so using the following command:

git lfs install

Cloning the Repository

To clone the entire repository into your local machine, use the following command:

git clone https://huggingface.co/ogrenenmakine/Refined-RadImagenet source/

This command will clone all files from the repository into a directory named source.

Notes

• Make sure you have sufficient storage space for large files.
• For more information about this dataset, visit the Hugging Face page.

Feel free to contribute or raise issues if you encounter any problems.

Usage

1. Extract the Dataset:

python process.py

Ensure the dataset tar file is located at:

source/

The script will automatically extract to:

data/

2. Process the Images: The script will read the CSV files, refined the images, and organize accordingly.

Dependencies

• Python 3.9+
• pandas
• OpenCV
• tarfile
• tqdm
• numpy

Install the required packages using pip:

pip install pandas opencv-python tarfile tqdm numpy

License

This project is licensed under the MIT License.