Automatic cell counting for phase‐contrast microscopic images based on a combination of Otsu and watershed segmentation method

High-content and high-throughput digital microscopes have generated large image sets in biological experiments and clinical practice. Automatic image analysis techniques, such as cell counting, are in high demand. Here, cell counting was treated as a regression problem using image features (phenotypes) extracted by deep learning models. Three deep convolutional neural network models were developed to regress image features to their cell counts in an end-to-end way. Theoretically, ensembling imaging phenotypes should have better representative ability than a single type of imaging phenotype. We implemented this idea by integrating two types of imaging phenotypes (dot density map and foreground mask) extracted by two autoencoders and regressing the ensembled imaging phenotypes to cell counts afterwards. Two publicly available datasets with synthetic microscopic images were used to train and test the proposed models. Root mean square error, mean absolute error, mean absolute percent error, and Pearson correlation were applied to evaluate the models’ performance. The well-trained models were also applied to predict the cancer cell counts of real microscopic images acquired in a biological experiment to evaluate the roles of two colorectal-cancer-related genes. The proposed model by ensembling deep imaging features showed better performance in terms of smaller errors and larger correlations than those based on a single type of imaging feature. Overall, all models’ predictions showed a high correlation with the true cell counts. The ensembling-based model integrated high-level imaging phenotypes to improve the estimation of cell counts from high-content and high-throughput microscopic images.

Download Full-text

CellCountCV—A Web-Application for Accurate Cell Counting and Automated Batch Processing of Microscopic Images Using Fully Convolutional Neural Networks

Sensors ◽

10.3390/s20133653 ◽

2020 ◽

Vol 20 (13) ◽

pp. 3653

Author(s):

Denis Antonets ◽

Nikolai Russkikh ◽

Antoine Sanchez ◽

Victoria Kovalenko ◽

Elvira Bairamova ◽

...

Keyword(s):

Neural Networks ◽

Endoplasmic Reticulum ◽

Endoplasmic Reticulum Stress ◽

Web Application ◽

Experimental Studies ◽

Living Cells ◽

Cell Counting ◽

Microscopic Images ◽

Cellular Models ◽

Cell Counts

In vitro cellular models are promising tools for studying normal and pathological conditions. One of their important applications is the development of genetically engineered biosensor systems to investigate, in real time, the processes occurring in living cells. At present, there are fluorescence, protein-based, sensory systems for detecting various substances in living cells (for example, hydrogen peroxide, ATP, Ca2+ etc.,) or for detecting processes such as endoplasmic reticulum stress. Such systems help to study the mechanisms underlying the pathogenic processes and diseases and to screen for potential therapeutic compounds. It is also necessary to develop new tools for the processing and analysis of obtained microimages. Here, we present our web-application CellCountCV for automation of microscopic cell images analysis, which is based on fully convolutional deep neural networks. This approach can efficiently deal with non-convex overlapping objects, that are virtually inseparable with conventional image processing methods. The cell counts predicted with CellCountCV were very close to expert estimates (the average error rate was < 4%). CellCountCV was used to analyze large series of microscopic images obtained in experimental studies and it was able to demonstrate endoplasmic reticulum stress development and to catch the dose-dependent effect of tunicamycin.

Download Full-text