Pytorch implementation of Label-Free Prediction of Cell Painting from Brightfield Images
We predict the Cell Painting image channels from a brightfield input using two models (U-Net and cWGAN-GP), and then extract the morphological features of interest. The model predictions are tested with a traditional segmentation-based feature-extraction approach, which allows us to explore evaluation methods of targeted biological relevance.
Note: we have not provided the image dataset due to AstraZeneca licenses but this may be available on reasonable request.
If you find this work useful, please consider citing our paper:
@article {10.1038/s41598-022-12914-x,
author = {Cross-Zamirski, Jan and Mouchet, Elizabeth and Williams, Guy and Sch{\"o}nlieb, Carola-Bibiane and Turkki, Riku and Wang, Yinhai},
title = {Label-Free Prediction of Cell Painting from Brightfield Images},
year = {2022},
doi = {https://doi.org/10.1038/s41598-022-12914-x},
URL = {https://www.nature.com/articles/s41598-022-12914-x#citeas},
journal = {Sci Rep 12, 10001}
}