Answer to Question #2461 Submitted to "Ask the Experts"

Q

I am a physics associate with a cancer treatment and research facility in Canada. Currently I am involved in a simple study to evaluate the percent error of LiF TLDs (thermoluminescent dosimeters) used to determine patient dose. The study involves using a small sample set of TLDs (48 = 1 box) from a larger population (240 in total) which has been sorted based on chip sensitivity. The sample set has been exposed to 200cGy using a cobalt-60 treatment unit. The TLDs were read and the exposure process repeated many more times.

Some questions have arisen based on my results. What is the uncertainty involved with using Harshaw TLD 100s (LiF) to determine dose? Where can I find information pertaining to the error of TLDs? Traditionally an error value of 5 percent has been quoted; how was this value determined? Given that a set of TLDs typically is sold in boxes of 48, what sort of statistics and reproducibility of dose should I be expecting? How much can I depend on a TLD to accurately measure dose? Will each chip consistently give a value of 200cGy if exposed to a known source of 200cGy (cobalt-60)? Plus or minus how many cGy? What sort of standard deviation should I be looking for? Has anyone else done this sort of simple study?

A

The use of LiF TLDs is common for verification of patient entrance dose. I used LiF discs and powders for such quality-control checks when I began my career in a large medical center over 25 years ago, using a manual Teledyne system. We used powders so we could blend them, thus didn't have to determine individual chip correction factors. You note in the question "a sort by sensitivity," but to improve precision, you might determine individual chip correction factors. A delivered dose of 200 cGy is typical for therapy, but quite large compared to an occupational measurement (i.e., 5 cSv/yr). And in my view, modern personnel dosimeters using TLDs should be accurate to +/-30 percent at this lower occupational dose level. ANSI N13.11-2001 (on the HPS Members Only website), provides a discussion of "tolerance levels" with testing personnel dosimeters.

Thus, I'd expect at 200 cGy you'd have a significant glow curve light output signal, with very good PMT (photomultiplier tube) current signal, and respective reproducibility. In order to calibrate your TLDs for accurate dose measurements, you are no doubt using a traceable ion chamber measurement for beam calibration. This beam calibration to some national primary standard has a certain error, e.g., a few percent. There are no doubt other factors to consider, e.g., timer error on delivered dose, electron contamination with unshielded TLDs, uniformity of the field if you're irradiating more than a single chip, free air beam/TLD calibrations vs. placement on a patient with backscatter. Repeated irradiations of the same chip to the same dose should provide a standard deviation on that chip's reproducibility. All these errors can be combined through proper propagation of error formula (see Knoll's text noted as a reference in this topical area of ATE, page 131). As far as the Harshaw TLD 100s (LiF) expected results, you should contact this manufacturer and request any application notes the manufacturer may have. As far as studies, I'm sure there are many, and a search of the American Academy of Physicists in Medicine's website and Medical Physics Journal would be a place to start.

You may want to check a few of the articles in the text Applied Thermoluminescence Dosimetry (Ed. By Oberhofer and Scharmann, available from All Bookstores.com). One article on the precision and accuracy of TLD measurements (on page 149), notes the precision for LiF at a 2 R exposure from 2 percent to 3 percent. Lastly, you may wish to discuss your results with the Radiological Physics Center (RPC) at the MD Anderson Hospital in Texas, as it provides a TLD QC service to the medical physics community.

David J. Allard, CHP