Automating dicentric chromosome detection from cytogenetic biodosimetry data

Area of Science:

• Biomedical image analysis
• Radiation biology
• Computational toxicology

Background:

• Accurate estimation of radiation exposure is critical for mass casualty incident response.
• Manual scoring of dicentric chromosomes (DCs) is time-consuming and subjective.
• Automated methods are needed to rapidly assess radiation doses in large populations.

Purpose of the Study:

• To develop and validate a prototype software system for automated dicentric chromosome (DC) counting.
• To achieve high throughput and accuracy in radiation exposure estimation.
• To enable rapid triage and medical management in mass casualty events.

Main Methods:

• Image segmentation using active contour gradient vector field (GVF) and discrete curve evolution (DCE).
• Centromere detection via minimizing width and DAPI-staining intensity profiles.
• Feature extraction for dicentric identification and algorithm validation for sensitivity and specificity.

Main Results:

• The algorithm achieved high sensitivity (85%) and specificity (94%) in dicentric chromosome identification.
• Image processing was accelerated significantly through parallelization (100-300x speedup).
• The software processed 200 metaphase images from 1025 specimens in 1.4 hours on a supercomputer.

Conclusions:

• The prototype software system demonstrates capacity and speed for estimating radiation exposures in mass casualty events.
• The automated approach is robust, independent of chromosome morphology and preparation protocols.
• This technology offers a significant advancement for biodosimetry in emergency situations.