What is the most suitable statistical method for comparing the results of two TLD systems and the dicentric assay? An intercomparison at dose ranges exceeding 100 mGy.

Thermoluminescent dosemeter (TLD) systems and dicentric chromosome assays (DCA) are crucial for accurate radiation dose estimation in emergencies. Ensuring that both techniques are well-calibrated and comparable is a key for safety in nuclear technology facilities. This study analysed different statistical methods to compare the responses of DCA and two TLD systems, (Harshaw 3500 and Harshaw 6600 Plus), for doses exceeding 100 mGy (up to 5Gy). Data were analysed using the determination coefficient (r2), Student's t-test, ANCOVA, and ISO 14146:2018. High correlations (r2 > 0.99) were obtained, ANCOVA revealing significant differences among systems (F = 7.671; P = 0.005; η2 = 0.006), although eta-squared indicated a minor relevance for these differences. According to ISO 14146:2018, all dosimetric systems showed similar results with overlapping confidence intervals. Based on these results, ISO 14146:2018 provided the most comprehensive comparison among dosimetry techniques, identifying deviations from optimal performance.