Predictive modelling of external dose rate from radionuclide therapy patients following discharge from hospital

Abstract

Background: Di erentiated thyroid carcinoma patients undergo thyroidectomy, followed by radiodine ablation in hypothyroid or euthyroid state. Several attempts were made to predict the external dose rate, however, there is a lack of accuracy in predictions from existing models of dose rate decay. Objectives: Published models were compared to actual patient measurements to test their validity. A patient speci c model was developed to improve the accuracy in predicting the external dose rate on the seventh day post ingestion, based on in patient measurements. This study also tackled the e ect that recombinant thyroid stimulating hormone has on the e ective half life of I131 in patients. Finally, this project tested a di erent way to model the external dose rate, using the gamma variate t. Research Design: Matlab was used to compare actual patient measurements to published model. Mathematica was used to model patient measurements and check for accuracy and numerical stability of the model. Unpaired t-test was used to compare the e ective half lives of hypothyroid and euthyroid patients. Results: Published model overestimated the external dose rates by a mean value of 10 Sv/h. P values of 0.23 and 0.93 show that there is no signi cant di erence in e ective half life between hypothyroid and euthyroid patients. The model developed in this study under estimates the external dose rate measurement on the seventh day by an average of 0.5 muSv/h. Gamma variate has slightly better accuracy than the other models. Conclusion: Published models overestimate the external dose rate measurements in ablation patients. The main causes are the inaccuracy in average e ective half lives in the model and the fact that the models are population speci c rather than patient speci c. Recommendations: Further to this study, one needs to include a higher number of external dose rate measurements to model the dose rate decay more accurately