Risk assessment of a new acceptance testing procedure for conventional linear accelerators.

PURPOSE: New techniques and materials have recently been developed to expedite the conventional linac acceptance testing procedure (Med Phys. 2017;22), which use the electronic portal imaging device (EPID) for data collection. This new procedure is designed to be more efficient and robust than the conventional approach. The purpose of this work was to perform a comparative risk assessment of the two acceptance testing procedures (ATPs). MATERIALS AND METHODS: Failure Modes and Effects Analysis was used to assess risks for both ATP approaches. Five domain experts (Medical Physicists) comprised the analysis team. The risk assessment method and ranking scales were adopted from the AAPM TG-100. The number of failure pathways and associated risk priority numbers (RPNs) for the two ATP approaches were compared. RPNs > 100 were considered high-priority failure modes. RESULTS: Fewer failure pathways were determined for the new ATP (ATPEPID ) compared to the conventional ATP (ATPconv ) resulting in a 44% difference (n = 233 vs. n = 534, respectively). There were also 35% fewer RPNs > 100 for the ATPEPID (n = 40) compared to the ATPconv (n = 114). Failure pathways and RPNs > 100 for individual ATP tests were 2.0 and 3.5 times higher, on average, for the ATPconv compared to the ATPEPID , respectively. The EPID pixel sensitivity map was identified as a high risk failure for the ATPEPID . CONCLUSIONS: Potential errors due to human factors were decreased for the ATPEPID compared to ATPconv so it is possible that a largely automated linac ATP can mitigate many error occurrences. Manufacturers should be careful when designing an EPID-based ATP to address errors in the EPID pixel sensitivity map which can potentially lead to a significant impact on patients' treatment.

Rapid acceptance testing of modern linac using on-board MV and kV imaging systems.

PURPOSE: The purpose of this study was to develop a novel process for using on-board MV and kV Electronic Portal Imaging Devices (EPIDs) to perform linac acceptance testing (AT) for two reasons: (a) to standardize the assessment of new equipment performance, and (b) to reduce the time to clinical use while reducing physicist workload. METHODS AND MATERIALS: In this study, Varian TrueBeam linacs equipped with amorphous silicon-based EPID (aS1000) were used. The conventional set of AT tests and tolerances were used as a baseline guide. A novel methodology was developed or adopted from published literature to perform as many tests as possible using the MV and kV EPIDs. The developer mode on Varian TrueBeam linacs was used to automate the process. In the EPID-based approach, most of mechanical tests were conducted by acquiring images through a custom phantom and software tools were developed for quantitative analysis to extract different performance parameters. The embedded steel-spheres in a custom phantom provided both visual and radiographic guidance for beam geometry testing. For photon beams, open field EPID images were used to extract inline/crossline profiles to verify the beam energy, flatness and symmetry. EPID images through a double wedge phantom were used for evaluating electron beam properties via diagonal profile. Testing was augmented with a commercial automated application (Machine Performance Check) which was used to perform several geometric accuracy tests such as gantry, collimator rotations, and couch rotations/translations. RESULTS: The developed process demonstrated that the tests, which required customer demonstration, were efficiently performed using EPIDs. The AT tests that were performed using EPIDs were fully automated using the developer mode on the Varian TrueBeam system, while some tests, such as the light field versus radiation field congruence, and collision interlock checks required user interaction. CONCLUSIONS: On-board imagers are quite suitable for both geometric and dosimetric testing of linac system involved in AT. Electronic format of the acquired data lends itself to benchmarking, transparency, as well as longitudinal use of AT data. While the tests were performed on a specific model of a linear accelerator, the proposed approach can be extended to other linacs.