How to Detect Stuff:  Critical Decision Levels and Minimal Detectable Limits


Alan W. Hunt
Idaho
Accelerator Center

In many health physics and physics experiments, the researcher needs to detect small quantities of stuff.  This stuff can be radioactive materials (i.e. our favorite example), chemicals (e.g. drugs) or elementary particles (e.g. leptons).  How do you make the critical decision to state that the stuff has been detected?  You probably have heard and/or used criteria like two times above background, two sigma above background and/or a signal to noise ratio of ten.  What do these statements mean and are they the best criteria for critical decision levels?  In this presentation, which will be part colloquium and part lecture, we will discuss what is actually being measured in a detection experiment, what is the best way to define (in my opinion) a critical decision level and how do you calculate your very own critical decision level.  Needless to say, we will focus the discussion on situations where Poisson distributions can be applied (e.g. radioactivity).  While critical decision levels are loads of fun, the minimal detectable limit, which is the smallest quantity of stuff that can be detected, is more important for understanding the limits of a detection system.  Of course, critical decision levels and minimal detectable limits are closely related and we will extend our discussion on critical decision levels to minimal detectable limits.  Throughout the presentation, I will use examples from my groups current research into fissionable material detection for security, safeguards and nuclear forensics applications.