Update on AAPM Task Group No. 43 Report--Brachytherapy and TLD

Abstract

Sir, The non-universality of many of the thermoluminescent (TL) characteristics of commonly used dosimetric materials, in general, and LiF:Mg,Ti, in particular, has been the subject of much discussion and controversy(1,2). Indeed, it appears to be one of the main disadvantages of thermoluminescent dosimetry (TLD). The following are the primary reasons for the lack of universality in the TL properties of peak 5 (the major glow peak used for dosimetric applications) in LiF:Mg,Ti. (1) In the preparation of the material, the TL trapping and luminescent centres are formed from the addition of impurities at the parts per million (ppm) level to the crystal-growth-melt (∼150 ppm w−1 Mg and ∼10 ppm w−1 Ti). Thus, other unwanted or stray impurities at the ppm level can and do create variations in the TL characteristics. (2) The lack of a universally and generally accepted experimental protocol controlling the various aspects of thermal and storage history, annealing, readout and analysis. These non-universalities can lead to very significant problems of interpretation which can be seriously exacerbated, especially when high precision and accuracy are required in the dosimetric application. An illustrative example is in the application of TLDs to low-energy X-ray interstitial brachytherapy using 125I (∼27.3 keV) and 103Pd (∼20.1 keV) X rays. One of the largest errors involved in this application as listed in the American Association of Physicists in Medicine (AAPM) Task Group No. 43 Report(3) is claimed to have a value of 5% (out of a total combined uncertainty of 8.7%) due to uncertainties in the experimental evaluation of the X-ray over-response. The over-response refers to an increased relative sensitivity to X rays <300 keV in energy and is ∼10% greater than that expected from the energy dependence of the mass–energy absorption coefficients. The phenomenon was probably first measured by Cameron et al.(4) 45 years ago, and the results of 10 independent experimental investigations for peaks 4 + 5 in LiF:Mg,Ti (TLD-100) were reviewed in detail 35 years ago(5). Between 20 and 30 keV, the over-response ‘for that part of the glow curve measured up to a maximum temperature of 240°C in the work of Budd et al.(5)’ appears to be 10–15%, but some early investigations actually reported an under-response at these low energies. A more modern compilation for LiF:MTS-N (LiF:Mg,Ti manufactured in Poland)(6) shows the same general features, i.e. an over-response of ∼10% at energies above 30 keV. But again, at 20 keV, the experimental data indicates an under-response of ∼10%. Two of the data points are 10% lower than what would be expected using the mass–energy absorption coefficients, one data point exactly agrees. These differences certainly arise in part (or in full) from the lack of a universally accepted protocol and have been the subject of fairly recent papers in the medical-physics dosimetry community attempting to understand the reasons behind the discrepancies in the various experimental measurements and Monte Carlo simulations(7). It deserves mention that the over-response is glow-peak-dependent; for the high-temperature thermoluminescence (HTTL: referring to thermoluminescence appearing at temperatures above ∼ 240°C), the over-response is ∼100%. Since the intensity of the HTTL relative to peak 5 is ∼10–20% at a dose level of 1 Gy, inclusion of even part of the HTTL in the TL signal may therefore influence the measured over-response of peaks 4 + 5 at the few per cent level. The relative intensity of the HTTL in the glow curve is critically dependent on the cooling rate following the 400°C pre-irradiation anneal as well as other material and batch-dependent factors(8,9). This might be an area of endeavour in which the application of computerised glow-curve deconvolution would decrease the discrepancies measured in various experimental investigations(10).