Related Experiment Video
Updated: Jan 3, 2026
07:57
A High-Throughput Comet Assay Approach for Assessing Cellular DNA Damage
Published on: May 10, 2022
5.8K
Classification of DNA damages on segmented comet assay images using convolutional neural network
Ümit Atila1, Yusuf Yargı Baydilli1, Eftal Sehirli2
1Department of Computer Engineering, Faculty of Engineering, Karabuk University, Karabuk, Turkey.
Computer Methods and Programs in Biomedicine
|November 17, 2019
Summary
Convolutional Neural Networks accurately classify DNA damage from comet assay images. This novel method achieves 96.1% accuracy, offering a robust alternative for biomedical research.
Area of Science:
- Biomedical Research
- Computational Biology
- Genetics
Background:
- DNA damage identification requires robust methods.
- The comet assay is a cost-effective DNA damage analysis technique.
- Current quantification methods need improvement.
Purpose of the Study:
- To evaluate Convolutional Neural Networks (CNNs) for DNA damage quantification using comet assay images.
- To compare CNN performance against existing methods.
- To introduce CNNs as a novel approach for comet assay image analysis.
Main Methods:
- Utilized 796 grayscale comet assay images (170x170 resolution), classified into 4 damage levels (G0-G3).
- Employed data augmentation on 796 images to create a training set of 9995 images.
- Trained a Convolutional Neural Network model on the augmented dataset.
Main Results:
- The CNN model achieved an overall accuracy of 96.1% in classifying comet images into 4 classes.
- The proposed CNN method is independent of image processing pre-processing parameters.
- Demonstrated high performance in classifying DNA damage levels.
Conclusions:
- Convolutional Neural Networks offer a novel and effective method for classifying comet assay images.
- This CNN approach provides accurate DNA damage quantification.
- Highlights the potential of AI in advancing comet assay analysis.
