Methods are described which relate the uncertainty in relative electron density derived from CT numbers to the uncertainty in treatment plan calculation for both photon and electron beams. These relationships are used to generate tolerance levels for electron density quality assurance measurements. These tolerance levels are dependent on treatment beam energy and tissue thickness, and are generally broader than current recommendations. The predicted treatment plan errors associated with these tolerance levels are shown to be consistent with calculations made using two commercial treatment planning systems. The tolerance levels are also shown to be practical by comparison against quality assurance measurements made on a conventional CT scanner and a treatment simulator CT system over a 12-month period. The results demonstrate that broader tolerances than are currently recommended can be justified in all situations except for electron beam treatments in which the therapeutic range falls within lung tissue.

Purpose: Having previously reviewed the implementation of systematic in vivo dosimetry at the Norfolk and Norwich Hospital this paper examines the results of entrance dose measurements for specific sites/techniques and determines whether different action/alert protocols are required for these different categories. Methods and materials: Entrance dose measurements using p-type diodes were analysed for the following treatment categories: Breast, head and neck in beam direction shell, abdomino-pelvic and intrathoracic. A 4% tolerance was applied. Results: Mean deviations from expected dose and proportion of measurements exceeding tolerance were: Breast: ϩ1.15% Ϯ 3.04% (1SD), 238/1073 Ն 4%; Head and neck: ϩ0.35% Ϯ 2.20% (1SD), 21/326 Ն 4%; Abdomino-pelvic: ϩ0.52% Ϯ 2.75% (1SD), 93/712 Ն 4%; Intrathoracic: Ϫ0.01% Ϯ 2.75% (1SD), 22/119 Ն 4%. Significant improvements in results for breast patients were noted following the introduction of a commercial breast board. The results for abdomino-pelvic patients confirmed a substantial variation in diode response under short FSD, wedged fields at 16 MV (that had not been corrected for). The statistical uncertainty in dose measurement for each treatment category was calculated in order to assist determination of appropriate tolerance levels. Conclusions: A blanket tolerance of 4% was generally too low given the extent of measurement uncertainty. The relatively high number of readings outside tolerance where identification of errors was difficult/impossible resulted in inconsistent application of the action protocol. Some widening of tolerances is likely to improve quality of procedure and treatment. Appropriate action levels are recommended for each treatment category. This value could be reduced if a variation in diode response under wedged fields at 16 MV was corrected for. § These values are larger mainly as a consequence of a relatively higher proportion of FSD inaccuracies for this category. Improvements in this respect would reduce these figures to the level seen for isocentric abdomino-pelvic techniques.

Purpose: This paper describes our experiences of implementing systematic in vivo dosimetry at the Norfolk and Norwich Hospital and reviews the results of 2,254 entrance dose measurements made over a 17-month period.Methods and materials: Entrance dose measurements using p-type diodes were performed on all new planned patients. The calibration procedure and correction factors applied are described. A 4% tolerance was applied.Results: The results of all measurements indicated a small mean deviation from expected entrance dose of ϩ0.77% and a standard deviation of 2.85%. 16.7% of all measurements exceeded the 4% tolerance with 9.2% exceeding a 5% level. The estimated overall errors for 578 treatments were calculated using the weighted averages of all beams. A narrower SD of 1.96% combined with only 4.8% of all treatments exceeding a 4% tolerance show that large deviations from a single field do not always translate into significant overall errors.Conclusions: Global dosimetric accuracy was within clinically acceptable limits and variations between measured and expected doses were mainly attributable to factors affecting diode reading. A number of errors in calculating deviations and the inconsistent application of the protocol suggest the need for interfacing the diode system with software control. KeywordsIn vivo dosimetry; diodes; quality assurance Systematic in vivo dosimetry for quality assurance using diodes

No abstract