Measurement theory in the clinic

Collecting data is one of the main day-to-day activities of the clinical physicist. Examples from the radiation therapy clinic include PDD's, mechanical alignment parameters, ambient temperature and pressure, source autoradiographs, IMRT planar dose distributions, setup SSD's, and CT number data. Patient outcomes, safety, and a host of other issues often rely on the quality and interpretation of that data.

One thing that sets medical physicists apart from most everyone else in the clinical environment and is crucial to performing our jobs well is having a strong grasp on the theory of measurement. Key concepts include: the statistical nature of data, the role of calibration, instrument resolution, and uncertainty and error propagation.

While we may not determine strict error levels for every measurement we make, it's important to have a grasp of these concepts while collecting and interpreting data.

• Why does my IMRT data taken with a planar detector array pass, even though it "looks" totally different than the TPS generated data? Probably because the detector array has relatively low resolution and the dose shape you see is the result of interpolation by software.
• Is my data skewed due to systematic or random errors?
• Is my new outlier subject to regression to the mean or the start of a new trend?
• Are my data groupings producing a Simpson's paradox?

What are your thoughts?