Code developed to perform interphase cell cycle staging of nuclei stained with DAPI. If you are using this code in your research please cite the paper.
To run this code please change the following directories in the file classifier.py:
-
img_dir:directory containing the DAPI images -
msk_dir:directory containing the segmentation masks corresponding to the DAPI images in directoryimg_dir(To obtain the segmentation masks for the DAPI images use the code available in https://github.com/HemaxiN/YOLO_UNET.) -
save_dir:directory where the results will be saved
To perform cell cycle staging run the file classifier.py after changing the img_dir, msk_dir and save_dir. After performing classification nuclei classified as S/G2 will have a green bounding box and nuclei classified as G1 will have a red bounding box in the images present in save_dir, as shown in the following figure:
Additionally, a results.csv file containing detailed information regarding nuclei classification will be saved in save_dir. It has the following structure:
| Image | pred_G1 | pred_S_G2 |
|---|---|---|
| image1.tif | 57 | 27 |
| image2.tif | 49 | 28 |
| ..... | ....... | ......... |
Each row contains the information for each image in img_dir. The first, second and third columns represent the image name, number of nuclei classified as G1 and number of nuclei classified as S/G2, respectively.
This implementation requires the packages listed in requirements.txt.
@article{narotamo2021machine,
title={A machine learning approach for single cell interphase cell cycle staging},
author={Narotamo, Hemaxi and Fernandes, Maria Sofia and Moreira, Ana Margarida and Melo, Soraia and Seruca, Raquel and Silveira, Margarida and Sanches, Jo{\~a}o Miguel},
journal={Scientific Reports},
volume={11},
number={1},
pages={1--13},
year={2021},
publisher={Nature Publishing Group}
}