Abstract book - ESTRO meets Asia

S113 ESTRO meets Asia 2018

Purpose or Objective In Queensland (Australia) a licencing condition to treat patients with radiation therapy is successful completion of a level III (whole process) audit conducted by an independent audit organisation. One such audit returned the following results: “Pass (Optimal)” for 3D conformal therapy, “Pass (Action Level)” for IMRT and “Out of Tolerance” for VMAT. An initial investigation revealed no obvious reason for the poor results. This paper presents a follow-up detailed and thorough investigation of the audit results. Material and Methods The level III audit involved irradiation of a modified version of the thorax phantom used in IAEA TecDoc 1583 (which describes tests for commissioning a treatment planning system). Doses were measured at a number of different points in the phantom for both homogeneous and inhomogeneous configurations. The audit tolerance limit was for the measured dose to be within 5% of the dose predicted by the treatment planning system. Differences in the derivation of kQ used by the auditor (measured) and the facility (tabulated data) were allowed for in the audit analysis. Most of the modulated audit cases were based on the C-Shape PTV and cylindrical avoidance core described by AAPM task group 119. Each step of the audit process was carefully examined and uncertainties were documented. Results All uncertainties found were within normally accepted and published levels (or manufacturer stated tolerances). It appears that for some of the audit points small uncertainties had added in the same direction to lead to an out of tolerance result for the VMAT audit. Some of these uncertainties were also present in the 3D conformal therapy audit which yielded a "Pass (Optimal)" result. The largest uncertainty came from the MLC leaf position accuracy with a 0.2 mm error in leaf calibration leading to dose discrepancies of up to 1.5% in the PTV core and more than 3% nearer the PTV peripherary. Uncertainties in an audit process which involved copying (small) structures representing ion chambers from a virtual auditor supplied dataset to a co-registered facility acquired CT under estimated the dose gradient present across the 6mm- diameter chamber by up to 4%. Conclusion Small uncertainties need to be considered carefully when performing modulated treatments. This audit provides evidence that dosimetric accuracy can be strongly affected by MLC leaf calibration accuracy. PO-275 Investigate of optimal structure setting for error detection on 3D dose verification system R. Nakahara 1 , R. Kawamorita 1 , K. Ishii 1 , S. Kishimoto 1 , K. Kubo 1 , Y. Nakasaka 1 , M. Kusawake 1 , T. Nakajima 1 1 Tane General Hospital, Radiation Oncology, Osaka, Japan Purpose or Objective The patients-specific quality assurance (QA) of the intensity modulated radiation therapy (IMRT) is performed not only to verify the dose distributions but also to detect error caused by human or machine. The COMPASS system (IBA Dosimetry) is three-dimensional dose verification system. It can validate using dose volume histogram, however there are a lot of parameters and indicators that it can be evaluated. The aim of this study is to evaluate the error detection capability when using COMPASS system in volumetric modulated arc therapy (VMAT) QA for localized prostate cancer and to investigate the optimal setting of structure for error detection. Material and Methods Ten patients with localized prostate cancer treated with VMAT in our institute were subjected. Twenty-five plans included following intentional error were generated,

which were energy change (10MV to 6MV), jaws retraction (5 mm, 3 cm) of one side and both sides, monitor unit change (+1%, +3%), systematic MLC open, close and shift error (0.2 mm, 0.35 mm, 0.5 mm) of each MLC side respectively. These plans were verified using COMPASS system, and no-error plan was verified to investigate false positive. A planning target volume (PTV), clinical target volume (CTV), organ at risk (OARs: rectum, bladder, right and left femoral head and penile bulb) and each percent of the prescribed dose volume (5, 10, 20, 30, 40, 50, 60, 70, 80, 90, 95, 98, 100: %PDV) were used as structure for evaluation. The Dmean, D98, D95 and D2 were used to evaluate PTV and CTV. The Dmean, V10, V20, V30, V40, V50, V60, V70, D2 were used to evaluate OARs and %PDV. Tolerance value was defined as average ± 3 standard deviation (SD) calculated in other 30 patients with localized prostate cancer treated by radiotherapy. The event was counted as error when each dosimetric indicators exceeded the tolerance value. Results The most sensitive dosimetric indicator was D2 of 5%PDV. The detection rate was 108/250 (error detection counts / all cases). PTV, CTV had high-sensitive parameters in error detection, otherwise sensitivity of OARs parameter was low. The 5%PDV and 70%-95%PDV had high-sensitive parameters in error detection, however middle dose %PDV had not high-sensitive parameters. The most sensitive structure was 5%PDV. The MLC shift error and 5 mm Jaws retraction were difficult to detect error and V70 of 90%PDV was the most sensitive structure for MLC close error detection. The result is summarized in table 1.

Conclusion To detect human error and machine error efficiently, it is important to understand how the error influence on the dosimetric indicators. The PTV and CTV had high sensitive parameters of error detection. Contrary, OARs had low. The percent of the prescribe dose structures had high sensitivity, especially 5%PDV, 80%PDV, 90%PDV were high. However, it was difficult to detect all type errors in a single structure. In this study, the combination of 5%PDV and 90%PDV was the highest detection capability structure setting.