Validation of semi-automatic scoring of dicentric chromosomes after simulation of three different irradiation scenarios.

Large scale radiological emergencies require high throughput techniques of biological dosimetry for population triage in order to identify individuals indicated for medical treatment. The dicentric assay is the "gold standard" technique for the performance of biological dosimetry, but it is very time consuming and needs well trained scorers. To increase the throughput of blood samples, semi-automation of dicentric scoring was investigated in the framework of the MULTIBIODOSE EU FP7 project, and dose effect curves were established in six biodosimetry laboratories. To validate these dose effect curves, blood samples from 33 healthy donors (>10 donors/scenario) were irradiated in vitro with 6°Co gamma rays simulating three different exposure scenarios: acute whole body, partial body, and protracted exposure, with three different doses for each scenario. All the blood samples were irradiated at Ghent University, Belgium, and then shipped blind coded to the participating laboratories. The blood samples were set up by each lab using their own standard protocols, and metaphase slides were prepared to validate the calibration curves established by semi-automatic dicentric scoring. In order to achieve this, 300 metaphases per sample were captured, and the doses were estimated using the newly formed dose effect curves. After acute uniform exposure, all laboratories were able to distinguish between 0 Gy, 0.5 Gy, 2.0, and 4.0 Gy (p < 0.001), and, in most cases, the dose estimates were within a range of ± 0.5 Gy of the given dose. After protracted exposure, all laboratories were able to distinguish between 1.0 Gy, 2.0 Gy, and 4.0 Gy (p < 0.001), and here also a large number of the dose estimates were within ± 0.5 Gy of the irradiation dose. After simulated partial body exposure, all laboratories were able to distinguish between 2.0 Gy, 4.0 Gy, and 6.0 Gy (p < 0.001). Overdispersion of the dicentric distribution enabled the detection of the partial body samples; however, this result was clearly dose-dependent. For partial body exposures, only a few dose estimates were in the range of ± 0.5 Gy of the given dose, but an improvement could be achieved with higher cell numbers. The new method of semi-automation of the dicentric assay was introduced successfully in a network of six laboratories. It is therefore concluded that this method can be used as a high-throughput screening tool in a large-scale radiation accident.

Automatic scoring of dicentric chromosomes as a tool in large scale radiation accidents.

Mass casualty scenarios of radiation exposure require high throughput biological dosimetry techniques for population triage in order to rapidly identify individuals who require clinical treatment. The manual dicentric assay is a highly suitable technique, but it is also very time consuming and requires well trained scorers. In the framework of the MULTIBIODOSE EU FP7 project, semi-automated dicentric scoring has been established in six European biodosimetry laboratories. Whole blood was irradiated with a Co-60 gamma source resulting in 8 different doses between 0 and 4.5Gy and then shipped to the six participating laboratories. To investigate two different scoring strategies, cell cultures were set up with short term (2-3h) or long term (24h) colcemid treatment. Three classifiers for automatic dicentric detection were applied, two of which were developed specifically for these two different culture techniques. The automation procedure included metaphase finding, capture of cells at high resolution and detection of dicentric candidates. The automatically detected dicentric candidates were then evaluated by a trained human scorer, which led to the term 'semi-automated' being applied to the analysis. The six participating laboratories established at least one semi-automated calibration curve each, using the appropriate classifier for their colcemid treatment time. There was no significant difference between the calibration curves established, regardless of the classifier used. The ratio of false positive to true positive dicentric candidates was dose dependent. The total staff effort required for analysing 150 metaphases using the semi-automated approach was 2 min as opposed to 60 min for manual scoring of 50 metaphases. Semi-automated dicentric scoring is a useful tool in a large scale radiation accident as it enables high throughput screening of samples for fast triage of potentially exposed individuals. Furthermore, the results from the participating laboratories were comparable which supports networking between laboratories for this assay.