@article{oai:repo.qst.go.jp:00047321,
 author = {Mano, Shuhei and 數藤, 由美子 and 數藤 由美子},
 issue = {4},
 journal = {Radiation and Environmental Biophysics},
 month = {Aug},
 note = {The dicentric chromosome assay (DCA) is one of the most sensitive and reliable
methods of inferring doses of radiation exposure in patients. In DCA, one calibration curve is prepared in advance by in vitro irradiation to blood samples from one or sometimes multiple healthy donors in considering possible inter-individual variability. Although the standard method has been demonstrated to be quite accurate for actual dose estimates, it cannot account for random effects, which come from such as the blood donor used to prepare the calibration curve, the radiation-exposed patient, and the examiners. To date, it is unknown how these random effects impact on the standard method of dose estimation. We propose a novel Bayesian hierarchical method that incorporates random effects into the dose estimation. To demonstrate dose estimation by the proposed method and to assess the impact of inter-individual variability in samples from multiple donors on the estimation, peripheral blood samples from 13 occupationally non-exposed, non-smoking, healthy individuals were collected and irradiated with gamma-rays. The results clearly showed significant inter-individual variability and the standard method using a sample from a single donor gave anticonservative confidence interval of the irradiated dose. In contrast, the Bayesian credible interval for irradiated dose calculated by the proposed method using samples from multiple donors properly covered the actual doses. Although the classical confidence interval of calibration curve with accounting inter-individual variability in samples from multiple donors was roughly coincident with the Bayesian credible interval, the proposed method has better reasoning and potential for extensions.},
 pages = {775--780},
 title = {A Bayesian hierarchical method to account for random effects in cytogenetic dosimetry based on calibration curves},
 volume = {53},
 year = {2014}
}