Gray-Level Co-occurrence Matrix Analysis for the Detection of Discrete, Ethanol-Induced, Structural Changes in Cell...

Area of Science:

• Microscopy and computational biology
• Cell biology and toxicology

Background:

• Gray-level co-occurrence matrix (GLCM) analysis is a powerful tool for assessing microscopic textural patterns.
• Understanding cellular responses to environmental stressors like ethyl alcohol is crucial in yeast biology.

Purpose of the Study:

• To apply GLCM analysis to Saccharomyces cerevisiae cell nuclei following sublethal ethyl alcohol damage.
• To evaluate the efficacy of machine learning (ML) models in distinguishing damaged from intact yeast cells based on GLCM features.

Main Methods:

• Calculated five GLCM parameters (angular second moment, inverse difference moment, contrast, correlation, textural variance) for each cell nucleus.
• Applied three ML models: neural network, random trees, and binomial logistic regression for classification.
• Analyzed GLCM features to identify differences between ethyl alcohol-treated and untreated yeast cells.

Main Results:

• Statistically significant differences in GLCM features were found between treated and untreated yeast cells.
• The multilayer perceptron neural network achieved the highest classification accuracy.
• The neural network model demonstrated high sensitivity, specificity, and discriminatory power for detecting alcohol-induced damage.

Conclusions:

• GLCM analysis combined with ML models can effectively detect alcohol-induced damage in Saccharomyces cerevisiae cell nuclei.
• This study presents the first GLCM-based ML model for sensitive detection of alcohol damage in yeast.
• The findings highlight the potential of computational microscopy for cellular toxicology assessment.