Automatic quantification of viability in epithelial cell cultures by texture analysis

Area of Science:

• Biomedical Imaging
• Cell Biology
• Computational Pathology

Background:

• Accurate quantification of live cells in microscopy is crucial for assessing cell viability in cultures.
• Manual counting of large cell populations is labor-intensive and prone to inaccuracies.
• Automated methods are needed for high-throughput and reliable cell viability assessment.

Purpose of the Study:

• To develop and validate an automated method for quantifying live and necrotic cells in phase contrast microscopy images using texture analysis.
• To compare the accuracy of the automated method with expert human assessment.
• To enable efficient and accurate in vivo assessment of cell viability in cultured cells.

Main Methods:

• Image segmentation based on texture analysis to classify regions into live cells, necrotic cells, and background.
• Utilized discriminant functions derived from histogram and co-occurrence matrix parameters.
• Employed stepwise discriminant analysis on a training set of 21 sample images.

Main Results:

• The developed method accurately quantifies the areas occupied by live and necrotic cells and the number of live cells.
• Results showed a very good correlation with expert observer assessment (Pearson's coefficient 0.95, kappa 0.87).
• The automated method allows for a significantly higher number of fields to be counted compared to manual methods.

Conclusions:

• An automated texture analysis method provides accurate quantification of cell viability in cultures.
• This method achieves expert-level accuracy while significantly increasing counting efficiency.
• The developed technique is valuable for in vivo assessment of cell viability in research and diagnostics.