Related Experiment Video
Updated: Feb 17, 2026
07:05
Area-based Image Analysis Algorithm for Quantification of Macrophage-fibroblast Cocultures
Published on: February 15, 2022
3.0K
Construction of a system using a deep learning algorithm to count cell numbers in nanoliter wells for viable
Takashi Kamatani1, Koichi Fukunaga2, Kaede Miyata3
1Pulmonary Division, Department of Medicine, Keio University School of Medicine, 35 Shinanomachi Shinjuku-ku, Tokyo 160-8582, Japan.
Scientific Reports
|December 6, 2017
Summary
A deep learning system using convolutional neural networks (CNNs) accurately counts cells in nanoliter wells for single-cell experiments. This advanced cell counting method achieves over 99% accuracy, improving high-throughput screening.
Area of Science:
- Biotechnology
- Computational Biology
- Microscopy
Background:
- Accurate cell counting is crucial for nanoliter well-based single-cell experiments.
- Traditional methods can be labor-intensive and prone to error.
Purpose of the Study:
- To develop and validate a deep learning-based system for automated cell counting in nanoliter wells.
- To compare the performance of the developed system against human expert determination.
Main Methods:
- A convolutional neural network (CNN) was trained on a large dataset of microscopic grayscale images (103,019 samples).
- The CNN classified wells into categories: 0, 1, 2, or more than 2 cells.
- An enhanced algorithm utilizing highest CNN outputs improved accuracy.
Main Results:
- The initial CNN system achieved 98.3% accuracy in classifying cell counts compared to human technicians.
- The enhanced system exceeded 99% accuracy.
- The system demonstrated robustness across different well shapes and outperformed other machine learning algorithms.
Conclusions:
- The developed CNN-based system provides a highly accurate and efficient method for cell counting in single-cell experiments.
- This automated approach is suitable for high-throughput and high-content screening applications.
- The system offers a reliable alternative to manual cell counting, especially in challenging imaging conditions.
