Center for Radiological Research, Columbia University Irving Medical Center, New York, NY, 10032, USA.
Radiological Research Accelerator Facility, Columbia University, Irvington, NY, 10533, USA.
Sci Rep. 2024 Aug 26;14(1):19789. doi: 10.1038/s41598-024-69243-4.
In major radiological events, rapid assays to detect ionizing radiation exposure are crucial for effective medical interventions. The purpose of these assays is twofold: to categorize affected individuals into groups for initial treatments, and to provide definitive dose estimates for continued care and epidemiology. However, existing high-throughput cytogenetic biodosimetry assays take about 3 days to yield results, which delays critical interventions. We have developed a multiwell-based variant of the chemical-induced G0-phase Premature Chromosome Condensation Assay that delivers same-day results. Our findings revealed that using a concentration of phosphatase inhibitor lower than recommended significantly increases the yield of cells with highly condensed chromosomes. These chromosomes exhibited increased fragmentation in a dose-dependent manner, enabling to quantify radiation damage using a custom Deep Learning algorithm. This algorithm demonstrated reasonable performance in categorizing doses into distinct treatment groups (84% and 80% accuracy for three and four iso-treatment dose bins, respectively) and showed reliability in determining the actual doses received (correlation coefficient of 0.879). This method is amendable to full automation and has the potential to address the need for same-day, high-throughput cytogenetic test for both dose categorization and dose reconstruction in large-scale radiation emergencies.
在重大放射学事件中,快速检测电离辐射暴露的方法对于有效的医疗干预至关重要。这些检测方法的目的有两个:将受影响的个体分为不同组别进行初步治疗,并为进一步的治疗和流行病学提供确切的剂量估计。然而,现有的高通量细胞遗传学生物剂量测定检测方法需要大约 3 天才能得出结果,这会延误关键的干预措施。我们开发了一种基于微孔板的化学诱导 G0 期染色体早熟凝聚检测方法的变体,可在同一天得出结果。我们的研究结果表明,使用低于推荐浓度的磷酸酶抑制剂可显著提高高度凝聚染色体的细胞产量。这些染色体表现出剂量依赖性的碎片化增加,使我们能够使用自定义的深度学习算法来定量评估辐射损伤。该算法在将剂量分为不同的治疗组方面表现出了合理的性能(对于三个和四个同剂量剂量箱,准确率分别为 84%和 80%),并且在确定实际接受的剂量方面具有可靠性(相关系数为 0.879)。这种方法易于实现完全自动化,有可能满足在大规模辐射紧急情况下对当天、高通量细胞遗传学检测的需求,用于剂量分类和剂量重建。