Minimum detectable activity, systematic uncertainties, and the ISO 11929 standard

The Currie formulation for minimum detectable activity (MDA) has served for decades as the standard method for estimating radiological detection limits-it is simple and statistically defendable. It does, however, lack a means to account for the effects of systematic uncertainties. In recent years we have seen various efforts to incorporate systematic uncertainties into an MDA framework. Perhaps most notable of these is the recent ISO standard 11929 for the determination of characteristic limits in ionizing radiation measurements. This standard brings a Bayesian perspective to the problem of characteristic limits in radiation measurements that are in many ways both welcome and long overdue. In this paper, however, we note some apparent drawbacks to the ISO 11929 approach. Namely, for small values of the systematic uncertainty the correction it makes to the Currie MDA is negligible, while for large systematic uncertainties the calculated MDA values can become infinite. In between these two extremes, the user has little basis for evaluating the reliability of the result. To address these issues, we consider the problem from a new approach, developing a straightforward phenomenological statistical model of the MDA that treats systematic uncertainties explicitly. We compare predictions from our model with results of the ISO 11929 formulation as well as the traditional Currie approach. Finally, some recommendations for alternative handling of the MDA in the face of significant systematic uncertainties are presented.