Portal Dosimetry (Varian Medical Systems, Palo Alto, USA) was the integrated patient specific QA system used in this project. It was used to acquire images of the measured fluence delivered from a Varian TrueBeam Platform version 2.7 while the Eclipse™ (v. 15.6.03, Varian Inc.) planning system was used to generate a modelled dosimetric predicted image. The verification plan was delivered directly to the EPID with the detector at an optimal distance of 100 cm from source. Dosimetric adaptation of the portal imager involved calibrating the acquired fluence to a CU to provide a form of absolute dose equivalence. The calibrated data was then compared to the predicted data attained through a modelled algorithm. The portal dose prediction calculation used the Portal Dose Image Prediction (PDIP) algorithm. Various QA tools for gamma analysis and point/profile dose were available with Aria RTM 16.1 Portal Dosimetry to determine plan validity . Portal dosimetry was used on a Varian TrueBeam™ linear accelerator with an aS1200-II EPID panel. Suitability of the panel for the HyperArc™ technique with flattening filter free (FFF) energy, lesion size and high number of monitor units was initially checked by its specifications . The detector panel had an active area of 40.0 x 40.0 cm2 with a pixel matrix of 1024 x 1024 pixels attaining a resolution of 0.0393 cm. Saturation occurred at dose rates above the maximum energy used and beam fluence was captured at SSD 100 cm. The fluence map collected, normalised to the center pixels of the panel, were calibrated to 1 CU to correspond to 100 MU at 100 cm SSD with 10 x 10 cm2 field for each energy . Constancy checks for output and uniformity were performed on a daily basis for each energy as per AAPM TG-142  report recommendation. Varian’s Machine Performance Check (MPC)  was used for these daily checks and verified beam constancy for PDIP verification plans.
The patient specific QA procedure for portal dosimetry required a verification plan to be created from the patient plan. Test cases were planned with Acuros (v 15.6.03) from which portal dose imaging prediction PDIP (v 15.6.03) verification plans were generated. The EPID had been commissioned for PDIP with checks covering variable dose rates, linearity, ghosting, output factors and intensity profile accuracy.
Film QA used both GafChromic EBT3 and EBT-XD with the latter used where the prescription dose exceeded 10 Gy . Films were exposed using LAP’s Easy Cube  where the modular inserts made it a useful tool for stereotactic dosimetry. The CIRS 30 cm x 30 cm slabs were initially trialled since they are commonly available in most physics departments and water equivalent however, due to their size excluded the use of the Encompass™ couch. The Easy Cube’s smaller size enabled it to be placed in the Encompass™ couch and its 18 cm sided cubic volume large enough to encompass the geometric range of intracranial mets. The Easy Cube’s polystyrene RW3 composition was not water equivalent, however the composition was taken into account by the Acuros TPS algorithm. The built-in radiopaque markers allowed for image guidance localisation. Films were scanned with an EPSON 1100 XL flatbed scanner following well documented QA procedures , , , , . A calibration file for each batch of EBT-XD and EBT3 GafChromic film was acquired using CIRS solid water slabs. The pixel value function  was dosimetrically validated with independent reference fields and rescaled during the patient QA checks with plan specific reference fields , .
The HyperArc™ patient verification plans were calculated with dose to water (Dw) by the Acuros algorithm with grid resolution and CT slice thickness both at 1 mm. Dw in transport medium was used with Acuros  since the film calibration was based on dose to water calculations. Coronal plane profiles were exported with a pixel resolution of 0.195 px/mm. Delivery of HyperArc™ patient plans in this project consisted of four arcs with 3 couch-kicks ± 45° and 90° from the sagittal plane. The kV/MV isocenter coincidence was checked through the Winston-Lutz (WL) test  before HyperArc™ delivery. AAPM TG 142  was referenced for the tolerances on the mechanical checks. Film was placed parallel to the coronal plane in the Easy Cube, marked with fiducials for laser alignment and then localised with a CBCT on a 6DOF couch. Any shift applied to the phantom was noted as this would affect the film/TPS registration in analysis. Every treatment plan delivery was accompanied by a calibration film as per protocol outlined by Lewis et al  to account for variation in machine daily output and film scanning conditions.
The scanned 48 bit 0.358 px/mm film images were analysed using Ashland’s FilmQAPro 2015 software. Each BM was initially analysed by comparing the FWHM and mean dose to validate dosimetric equivalence for the mets at various off-axis (OA) distances. Gamma analysis was then acquired for each met with gamma pass rate criteria set at dose difference of 2%, distance to agreement (DTA) set to 2 mm with a 10% threshold. Selection of the region of interest (ROI) was localised to the surrounding area of the BM.
The methods outlined in this section cover validation of film and PDIP dosimetry for multi-mets, geometrical and dosimetric uncertainties in the QA process for single and multi-met HyperArc™ analysis and BM case studies.
Although film and PDIP patient specific QA procedures were already established in the department, validation of a higher dose range for SRS mets and scanning of multi-mets was required. SRS plans generated maximum dose in excess of fractionated SRT plans and required extended dosimetric validation of calibration files for both EBT-XD and EBT3. Known issues as addressed by Chen et al  regarding variable scanning response in the lateral direction was potentially an issue for scanning multiple mets on a single film. Checks were done by scanning a range of exposed film for both EBT3 and XD over a range of 50 mm from the central longitudinal axis of the EPSON 1100 XL flatbed scanner. The ROI measuring the pixel values (PV) for each film strip were measured for both red and green channels. Percentage differences between the averaged 50 mm lateral off-set PV were compared to the central axis PV. Based on these results, recommendations were made for scanning films with multiple off-axis mets and the selection of appropriate colour channel for all other measurements pertaining to the remainder of this project.
A range of tolerances and criteria were investigated by Miften et al.,  for SRT/ SRS techniques from which dosimetric and geometric tolerances were nominally set 2%/2mm. These were selected to test patient QA processes, provide baselines for future development and were within the scope of machine deliverability. Film QA processes required investigating to determine the sources of greatest uncertainty due to inclusion of extended SRS dose range and multi-met film exposures. A film calibration curve model will vary from data measured due to the nature of curve fitting  and may vary at key points by greater than 2%. Variances in machine output, scanner temperature and film development time all caused further dosimetric uncertainty however, these uncertainties were minimised by the inclusion of reference films . Checks were performed to determine maximum variation of a range of exposed films of known dose over the range of the film calibration. Checks were performed to determine the geometric shifts required in the film QA gamma optimisation process for multi-mets. WL checks were recorded over period of 8 months with the requirement that tests would be repeated if a maximum delta shift was above 1 mm.
Single met plans were created using the HyperArc™ planning technique for mets 10–25 mm diameter at the isocenter with energies 6MV, 6FFF, and 10FFF. Each plan was prescribed to a single dose of 10 Gy with each plan using four non-coplanar arcs. Gamma values were acquired from both PDIP and Film QA Pro 2015. FWHM comparisons were measured for film and compared with the TPS. The FWHM was not a relevant measure with PDIP due to its 2D integral fluence. Based on the outcome of these results, target volume (TV) size limits would be recommended for both PDIP and film for the remainder of this project.
Multi-met plans were created with HyperArc™ for 20 mm TVs spaced at diagonal distances of 10 mm, 20 mm, 50 mm and 70 mm (see Fig. 2) from the isocentre. The TVs were placed diagonally so that OA distances in both in-plane and cross-plane axes were investigated. Each TV in each single fraction plan had a prescription dose of 10 Gy. The diagonal OA met plans were delivered using 6MV, 6MV FFF, and 10 MV FFF energies. Checks were performed to determine if there was any dosimetric dependency on distance of the TV from the iscocenter. PDIP and film gamma analysis were acquired based on ROI relative to the TV with 2%/2mm criteria. FWHM and mean dose was measured for each TV for film and compared to the TPS plan. The FWHM was averaged over the in-plane and cross-plane measurements.
Based on these results, met OAD limits and beam energy would be recommended for both PDIP and film.
Case studies were sampled from a set of previously treated BM patients from a sister site which used BrainLAB for their TPS. The plans were anonymised and re-planned using HyperArc™. From the range of eleven case studies, four were selected to represent a range of possible multi-met scenarios. Having determined patient specific QA guidelines on size limits for TVs, off-axis geometry, each of the four representative plans were initially screened against these guidelines. For each plan, this included an analysis of the size and location of each lesion, its proximity to organs at risk such as brain stem, optic chiasm and proximity to bone and to other lesions and lesion morphology. Mets not able to be delineated by the width of a single 5 mm pair of MLC leaves were planned as a single TV. Film QA was measured in the coronal plane of the TV’s near maximum dose. All of the selected test cases presented with at least seven lesions, making the chance of any one coronal plane intersecting more than one lesion high. In these cases, flexibility in coronal plane positioning was optimised to near maximum dose in multiple lesions. The phantom was setup in the Encompass™ couch and registered using CBCT. Due to limited number of slab combinations, some plans were required to be delivered more than once to have films placed in the correct coronal plane.