Abstrakt
Time evolution of the cumulated activity in human body is one of the key characteristics determining medical impacts of ionizing radiation. In nuclear medicine, so called effective half-life is mostly used for describing the evolution.
This quantity is usually estimated by fitting a straight line in semi-logarithmic coordinates. Its novel Bayesian estimate was also proposed and its advantageous properties were verified.
During extensive tests, it was found that the effective half-life has limited use as the underlying deterministic relationship time-activity can hardly be taken as (mono)exponential. It stimulated the search for a better and still simple model.
A quadratic dependence of in (activity) on in (time) was found as an adequate candidate. Preliminary experiments on a restricted set of real data were promising enough to justify its further elaboration.
The paper reports on the progress made in verifying and exploiting this non-standard model. Its potential exploitation in dosimetric tasks is outlined.
The core of the paper describes a positive verification of our results on an extensive set of real data.