Improving the accuracy of TLD systems in clinical applications

Conditions are outlined for obtaining routine measurements of the highest precision with a TLD system in clinical applications. A method for studying and specifying the inherent uncertainties in a TLD system is also presented. The principles of optimising reader design are given, as is guidance in optimising partial processes of the measurement (e.g. during annealing or constant dose irradiation). Finally, two alternative TLD processes and a clear-cut calibration procedure are presented, along with instructions for the operator on how to keep systematic errors at an insignificant level. The overall uncertainty of measurement using sintered LiF discs was compared with the required accuracy (to within 3%) for the calibration of therapy units. The conclusion was drawn that, to meet this requirement, the random error of measurements should be very small (1 SD below 0.5%) as it is in the best manual TLD systems. Automatic TL readers seem to be far from this goal, but improvements may be made by redesigning light collection geometries for the specific needs of clinical applications. (Author)