Agreement criteria between expected and measured field fluences in IMRT of head and neck cancer: The importance and use of the γ histograms statistical analysis

Background and purposeThe aim of this study was to analyse our pre-treatment QA data in order to establish uniquely defined agreement criteria between planned and delivered dose distribution for clinical QA practice in IMRT of head and neck (HN) patients.Materials and methodsPre-treatment QA dosimetry using films in combination with ionisation chambers is routinely evaluated for each patient. This evaluation is performed by comparing planned and measured dose distributions in terms of absolute point dose measurements, planar dose verification and γ function analysis using 4%/3 mm values as acceptance criteria. In the current investigation, γ histograms, calculated on dose levels higher than 10% of the prescription dose (1.8 Gy/fraction), were further analysed by considering mean values, γ values corresponding to Δ = mean + 1.5 SD (named as γΔ) and the % of points with γ < 1, γ < 1.5 and γ > 2. When considering the patient population, the average values of all these parameters and their confidence limits (mean value + 1.5 SD) were calculated. The results here presented refer to 57 HN patients treated in the period September 04–April 06.ResultsBetter results were found for treatments performed with our newly installed linac (e.g. average γΔ: 0.8 vs. 1.1 for the preexisting one, p < 0.001), due to a more accurate dosimetric configuration. Also, confidence limits for the percentage of points with γ < 1, γ < 1.5 and γ > 2 were found to be quite different for the two linacs (95.3%, 98.9% and 0.4% for the newly installed unit, 87.6%, 95.2% and 2.0% for the preexisting one).ConclusionsStatistical analyses of γ evaluation of QA pre-treatment dosimetry are useful to properly define confidence limits of the agreement between expected and measured fluences based on our institutional experience. Our results confirm that the dosimetry configuration of the beam may significantly affect the agreement between planned and measured IMRT beam fluences.