A Framework for Automatic Recognition of Cell Damage on Microscopic Images using Artificial Neural Networks.

Area of Science:

• * Cell biology
• * Computational biology
• * Toxicology

Background:

• * Traditional microscopy for cell damage assessment is laborious and time-consuming.
• * Digital Image Processing (DIP) offers potential for automating microscopic examinations.
• * Existing methods lack efficient tools for analyzing specific cellular damage patterns.

Purpose of the Study:

• * To develop and validate an automated framework for classifying animal cell damage.
• * To leverage DIP and machine learning for analyzing DNA damage in Helix aspersa cells.
• * To improve the efficiency and accuracy of microscopic slide examination.

Main Methods:

• * A framework integrating image segmentation, feature extraction, and a Shallow Neural Network (SNN).
• * Application of the Single Cell Gel Electrophoresis (SCGE) assay on Helix aspersa cells.
• * Training and validation using a dataset of 130 manually segmented and labeled slide samples.

Main Results:

• * The proposed framework achieved an average accuracy of 88.3% in classifying 5 types of animal cell damage.
• * Demonstrated robustness in analyzing DNA damage using the SCGE assay.
• * Successfully classified cell damage in Helix aspersa, a land mollusk model.

Conclusions:

• * The developed framework offers a robust and efficient solution for automated cell damage classification.
• * This approach can significantly aid professionals in microscopic examinations, reducing manual labor.
• * The study highlights the potential of integrating DIP and SNNs for toxicological assessments.