Radiation dosimetry by automatic image analysis of dicentric chromosomes

Area of Science:

• Cytogenetics
• Radiation Biology
• Medical Imaging

Background:

• Dicentric chromosome scoring is crucial for biological dosimetry after radiation exposure.
• Manual scoring is labor-intensive and prone to inter-observer variability.
• Automated systems are needed to improve efficiency and accuracy.

Purpose of the Study:

• To develop and validate a fully automated image analysis system for scoring dicentric chromosomes.
• To establish a reliable dose-response curve for gamma irradiation using the automated system.
• To assess the system's efficiency and accuracy compared to manual analysis.

Main Methods:

• The system automates cell detection, chromosome segmentation, and centromere location.
• A calibration set of 15,000 cells established the dose-response curve for 60Co gamma irradiation.
• Maximum likelihood and Monte Carlo techniques were used for parameter estimation and confidence limits.
• A blind trial with 8000 cells validated the system's dose estimation accuracy.

Main Results:

• The automated system demonstrated a close correspondence between estimated and true radiation doses.
• The system achieved approximately 40% sensitivity in detecting dicentric chromosomes.
• Analysis of sufficient cells for equivalent visual analysis of 500 cells required about 1 hour of operator review.

Conclusions:

• The automated image analysis system provides an efficient and accurate method for dicentric chromosome scoring.
• The system facilitates reliable biological dosimetry by establishing a robust dose-response relationship.
• While requiring analysis of more cells, the automated system offers significant time savings for operator review.