Efficient biomarkers for the characterization of bone tissue

Area of Science:

• Biomedical Engineering
• Computational Biology
• Histopathology

Background:

• Accurate classification of cell tissue is crucial for understanding bone tissue regeneration from stem cells.
• Microscopic image analysis presents challenges due to complex phenotypes and color variations.

Purpose of the Study:

• To develop and evaluate an expert system for accurate cell tissue classification in bone regeneration studies.
• To identify optimal feature extraction methods and classifiers for distinguishing bone tissue from other cell types.

Main Methods:

• Feature extraction using texture, shape, and color descriptors (histograms, Zernike moments, circular parameters).
• Analysis of various classifiers including neural networks, decision trees, Bayesian classifiers, and association rules.
• Image decomposition into smaller windows to improve execution speed and robustness.

Main Results:

• Safranin blue staining demonstrated superior performance compared to Picrosirius red and alcian blue.
• The multilayer perceptron classifier achieved the highest accuracy in distinguishing bone tissue.
• Image window decomposition positively impacted execution speed without significant loss of robustness.

Conclusions:

• The developed expert system effectively classifies cell tissue in bone regeneration images.
• Safranin blue staining and multilayer perceptron offer a robust and accurate approach for histological analysis in this field.