DeepFoci: Deep learning-based algorithm for fast automatic analysis of DNA double-strand break ionizing...

Area of Science:

• Biophysics
• Molecular Biology
• Radiology

Background:

• DNA double-strand breaks (DSBs) are critical DNA lesions.
• Ionizing radiation-induced foci (IRIFs) are key markers for DSBs.
• Accurate IRIF quantification is essential for radiation biodosimetry and understanding DNA repair.

Purpose of the Study:

• To develop and validate DeepFoci, a fully automatic deep learning method for IRIF counting and morphometric analysis.
• To improve the accuracy and efficiency of DSB detection and repair studies.

Main Methods:

• DeepFoci utilizes U-Net for nucleus segmentation and IRIF detection in 3D multichannel data.
• It incorporates maximally stable extremal region-based segmentation for IRIF identification.
• The method was trained and tested on diverse cell types and irradiation conditions.

Main Results:

• DeepFoci achieved the highest accuracy in IRIF quantification compared to existing algorithms.
• Its detection error was comparable to expert variability across a wide range of IRIF counts.
• The software extracted 3D morphometric features and analyzed protein colocalization within IRIFs.

Conclusions:

• DeepFoci offers a robust and accurate solution for IRIF quantification.
• It refines the analysis of DNA repair processes and patient tumor classification.
• The method has broad applications in radiotherapy monitoring, biodosimetry, and understanding radiation-induced DNA damage.