# CNN Hands-on Practice — Data

来源：https://www.kaggle.com/competitions/cnn-hands-on-practice/data

本地 Meta Kaggle 快照；不是本次实时页面。

# Dataset Description

## Context

The whole world is suffering from COVID-19 disease and its proper and timely diagnosis is the need of the hour. So for building an efficient AI-based diagnostic system we have collected Chest ray images from different sources so that we can train our CNN model for automating the whole diagnosis method. Therefore we have collected Chest ray images from different sources and research papers and combined them to create one comprehensive dataset that can be used by research community.

This dataset is also used in COVID Lite paper which has shown significant results by building novel CNN based solution.

## Content

This dataset consists of a posteroanterior (PA) view of chest X-ray images comprising Normal, Viral, and CVOID-19 affected patients. There are total 1823 CXR images.

## Acknowledgements

Please cite these papers if you are using it for any scientific purpose:

1. D. Kermany, K. Zhang, M. Goldbaum, Large dataset of labeled optical coherence tomography (oct) and chest x-ray images, Mendeley Data, v3 http://dx. doi. org/10.17632/rscbjbr9sj 3 (2018).
2. C. J. P, P. Morrison, D. L, Covid-19 image data collection, arxiv, arXiv preprint arXiv:2003.11597 (2020). URL https://github.com/ieee8023/covid-chestxray-dataset
3. Z.-H. Chen, Mask-rcnn detection of covid-19 pneumonia symptoms by employing stacked autoencoders in deep unsupervised learning on low-dose
    high resolution ct (2020). doi:10.21227/4kcm-m312.URL http://dx.doi.org/10.21227/4kcm-m312
4. D. S. et.al, Covid19action-radiology-cxr (2020). doi:10.21227/s7pw-jr18.
    URL http://dx.doi.org/10.21227/s7pw-jr18