Session Item

In vivo dosimetry (IVD) is known to be an effective quality assurance method to detect deviations between prescribed and administrated dose. From the available IVD systems, the Electronic portal imaging device (EPID) has been shown to have great potential for large scale acquisitions of in vivo dosimetry (IVD) data. In recent years, many institutes developed EPID IVD programs with commercial or homemade software. In this work we investigated the critical aspects and the effectiveness of the IVD implementation with a focus on EPID based IVD on an international level.

An ESTRO working group, created a survey containing 44 questions [Fig1] divided in three parts: IVD software characteristics, software implementation and data analysis, and patients results. The survey was sent to the ESTRO mailing list on 14 October 2019 and was closed on 30 November 2019.

133 institutes participated in the survey, of which 45 used IVD with EPID. 17 centres submitted statistics of patient IVD data. The software were: Perfraction (SunNuclear) (9), Epigray (DOSISsoft) (8), Softdiso (Best Medical) (7), Dosimetry Check (LAP) (6), Portal Dosimetry (Varian) (6), homemade software (6), and IviewDose (Elekta) (3). These in vivo solutions involve different degrees of automation as well as different metrics for dose comparison. An average of 2657 patients (range 50-8000) and 10021 fractions (range 200-24000) were annually analyzed with fully automatic software, compared to 855 patients (range 100-2500) and 2791 fractions (range 1000-6575) with the partially automatic and 266 patients (range 20-1200) and 378 fractions (range 30-3500) with the software without automation. The percentage of all measurements with results out of institutional tolerance limits, was, on average, 14% with 10% standard deviation. The causes of out of tolerance results were: patient anatomical variations 5.2%, setup errors 3.7%, incorrect use of immobilization devices 2%, beam delivery discrepancies 1.3% and incorrect plan computation 0.3%. Failure of the IVD algorithm or erroneous EPID acquisitions occurred in 4.5% of fractions. In 2.2% of fractions, the cause was not unidentified. Sporadic network data transfer errors were reported by one centre. The actions taken after out of tolerance detection, were, on average: repetition of the IVD test 8.5%, instruction of the technician for improving the patient setup 3.8%, informing the radiation oncologist 0.9%, and replanning 1.6%.

In this study, the degree of global implementation, the high potential to detect and correct errors, and the challenges for wider spread use of EPID IVD have been shown. However, the results are very heterogeneous as implementation, data analysis and underlying causes for out of tolerance are centre dependent. International guidelines could facilitate harmonization and standardization of EPID IVD use and facilitate efficient implementation.