Purpose: To discriminate between random and systematic treatment setup errors using portal films.
Methods and materials: A bi-dimensional analytic techniques using multiple analyses of variance based on Hotelling's T2 statistics to derive numerical and graphical measures of daily portal film accuracy and precision has been trialed using 88 daily portal films from seven patients' treatment.
Results: A demonstration is provided of how a reasonable approximation of random variation from the intended (Simulator) field center, and systematic displacement of the mean position of the portal film centers may be derived from a minimum number of portal films. If a random error as great as 10 mm exists, at least six or seven portal films are considered necessary to reliably detect and quantify the size of a systematic error.
Conclusion: Our results suggest that a modest systematic error could go undetected until the end of a 5 or 6 week course of treatment if only one portal film is obtained each week. A greater number of portal films should be performed during the first week of treatment to reduce the frequency of such errors. Efforts to separate and quantify both random and systematic errors in setup are worthwhile and will lead to improvements in outcome at the individual patient level and at a departmental level in the development of quality assurance programs.