Pseudo-nuclear staining of cells by deep learning improves the accuracy of automated cell counting in a label-free cellular population

https://doi.org/10.1016/j.jbiosc.2020.09.014Get rights and content

Deep learning has emerged as a breakthrough tool for the segmentation of images without supporting human experts. Here, we propose an automated approach that uses deep learning to generate pseudo-nuclear staining of cells from phase contrast images. Our proposed approach, which has the feature to generate pseudo-nuclear stained images by simple DNN, showed remarkable advantages over existing approaches in the cell-detection and the detection of the relative position of cells for various cell densities, as well as in counting the exact cell numbers. Pseudo-nuclear staining of cells by deep learning will improve the accuracy of automated cell counting in a label-free cellular population on phase contrast images.

Section snippets

Cell line and construction of m5s cells with nucleus-localized RFP

Fibroblast-like m5s mouse embryonic cells were obtained from JCRB Cell Bank, National Institutes of Biomedical Innovation, Health and Nutrition (Tokyo, Japan). The pmCherry-H3 plasmid DNA (19) was transfected into m5s cells (20) and these were screened with G418 to obtain stable m5S-RFP cells with mCherry-labeled nuclei.

Cell culture

Stable m5S-RFP cells (5 × 104 cells) were seeded on a 35 mm dish and grown in DMEM (4.5 g/L glucose; Nacalai Tesque, Kyoto, Japan) supplemented with 10% fetal bovine serum (ICN

Comparison of two types of processed images generated from phase contrast images of m5s-RFP cells

To compare pseudo-nuclear stained images generated by our DNN algorithm with images restored by a previously reported deconvolution algorithm (8), original 8-bit grayscale images were processed by the both algorithms. Representative images at 24 h, 72 h, and 120 h after seeding of m5s-RFP cells are shown in two columns on the left side of Fig. 3A. As a control, original fluorescence and phase contrast images are shown in two columns on the right side of the same figure.

Extracted pseudo-nuclear

Acknowledgments

We thank Ai Kawakita-Yamaguchi (Live Cell Imaging Institute, Osaka Japan) for construction of m5S-RFP cells. This study was performed by collaboration among AIST, Dai Nippon Printing Co., Ltd. and Osaka Prefecture University. AIST was supported by Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Number JP17H03472. This study was funded by Dai Nippon Printing Co., Ltd. Tsuzuki and Sanami are employees of Dai Nippon Printing Co., Ltd.

S.S. and S.F. designed the experiments. S.F.

References (20)

There are more references available in the full text version of this article.
View full text
© 2020 The Society for Biotechnology, Japan. All rights reserved.