Application of an imaging flow cytometry γ-H2AX assay for biodosimetry using supervised machine learning

Area of Science:

• Biomarkers
• Radiation Biology
• Cellular Assays

Background:

• Phosphorylation of histone H2AX (γ-H2AX) is a rapid indicator of DNA double-strand breaks (DSBs) caused by ionizing radiation.
• γ-H2AX serves as a biomarker for radiation exposure, traditionally measured by microscopy or flow cytometry.
• Imaging flow cytometry (IFC) merges microscopy resolution with flow cytometry's statistical power for γ-H2AX analysis.

Purpose of the Study:

• Develop and validate an imaging flow cytometry (IFC) assay for measuring γ-H2AX in human lymphocytes.
• Optimize sample acquisition and data analysis for accurate radiation dose estimation.
• Compare the accuracy of supervised machine learning (SML) K-Nearest Neighbors (K-NN) algorithm with traditional methods for dose assessment.

Main Methods:

• An IFC method was developed to measure γ-H2AX in human lymphocytes.
• The assay was optimized and validated for the 0-10 Gy dose range.
• Supervised machine learning (SML) K-Nearest Neighbors (K-NN) algorithm utilized mean fluorescence intensity (MFI), spot count, and spot area for dose estimation.

Main Results:

• The SML K-NN method demonstrated higher accuracy in determining delivered radiation doses compared to MFI-based linear fitting.
• This improved accuracy was particularly notable in the higher dose range (7-10 Gy).
• The γ-H2AX-IFC assay showed enhanced efficiency 1 hour post-exposure.

Conclusions:

• The validated γ-H2AX-IFC assay, coupled with SML K-NN for dose estimation, offers an efficient method for assessing radiation exposure.
• This approach improves upon traditional flow cytometry dose estimation techniques.
• The assay has potential as a rapid triage tool for screening large sample numbers after radiation incidents.