Stereotactic Optimized Automated Radiotherapy (SOAR): a novel automated planning solution for multi-metastatic SRS compared to HyperArc™

Objective. Automated Stereotactic Radiosurgery (SRS) planning solutions improve clinical efficiency and reduce treatment plan variability. Available commercial solutions employ a template-based strategy that may not be optimal for all SRS patients. This study compares a novel beam angle optimized Volumetric Modulated Arc Therapy (VMAT) planning solution for multi-metastatic SRS to the commercial solution HyperArc. Approach. Stereotactic Optimized Automated Radiotherapy (SOAR) performs automated plan creation by combining collision prediction, beam angle optimization, and dose optimization to produce individualized high-quality SRS plans using Eclipse Scripting. In this retrospective study 50 patients were planned using SOAR and HyperArc. Assessed dose metrics included the Conformity Index (CI), Gradient Index (GI), and doses to organs-at-risk. Complexity metrics evaluated the modulation, gantry speed, and dose rate complexity. Plan dosimetric quality, and complexity were compared using double-sided Wilcoxon signed rank tests (α = 0.05) adjusted for multiple comparisons. Main Results. The median target CI was 0.82 with SOAR and 0.79 with HyperArc (p < .001). Median GI was 1.85 for SOAR and 1.68 for HyperArc (p < .001). The median V12Gy normal brain volume for SOAR and HyperArc were 7.76 cm3 and 7.47 cm3 respectively. Median doses to the eyes, lens, optic nerves, and optic chiasm were statistically significant favoring SOAR. The SOAR algorithm scored lower for all complexity metrics assessed. Significance. In-house developed automated planning solutions are a viable alternative to commercial solutions. SOAR designs high-quality patient-specific SRS plans with a greater degree of versatility than template-based methods.


Introduction
Stereotactic Radiosurgery (SRS) is a well-established treatment technique for benign and malignant brain lesions using a single high dose of radiation.For large target volumes or targets close to an organ at risk Fractionated Stereotactic Radiotherapy (FSRT) in three or five fractions may be the preferred treatment.The use of SRS and FSRT in the management of brain metastases has increased based on evidence from clinical trials that have shown improved cognitive function with SRS compared to Whole Brain Radiation Therapy (WBRT) and the non-inferiority of SRS for the treatment of 5 to 10 metastases compared to 2 to 4 metastases (Yamamoto et al 2014, Brown et al 2016).Data from the Central Cancer Registry showed that SRS was used as the initial form of treatment for 15% of patients with 1 to 3 metastases in 2007-2011rising to 68% in 2017(Batra et al 2021)).For patients with 4 to 9 lesions this rate jumped from 3% to 33% (Batra et al 2021).Multi-metastatic SRS treatment planning is a complex and time-consuming process with multiple targets, often close to Organs at Risk (OARs).Treatment goals include preserving cognitive function by minimizing dose to OARs within the cranium and achieving adequate coverage of all target volumes.With SRS replacing WBRT as the standard of care for intra-cranial metastases, the number and complexity of SRS treatment plans has increased, putting a strain on radiation oncology departmental resources.Automated treatment planning solutions have been introduced with promises of improved efficiency and plan quality by reducing dosimetric variability.These include Varian's HyperArc™ (Varian Medical Systems, Palo Alto, California) and Brainlab Elements™ (Brainlab, Munich, Germany) (Boczkowski et al 2020).These commercial solutions demonstrate reasonable plan quality and efficiency, however, include downsides such as additional costs for software licenses, strict immobilization system requirements, and reliance on template-based planning that may not be suitable for all patients.
The introduction of scripting in radiation oncology has given clinics the ability to develop custom software integrated directly with the treatment planning system.This software can be tailored to meet institutional processes and guidelines.At the Tom Baker Cancer Centre an automated SRS planning solution was developed to individualize patient treatment by optimizing beam trajectory placement on the allowable collision free space around the patient (Mann et al 2019), and validated on a retrospective cohort of 25 patients (Mann, Thind and Ploquin, 2022).This solution was clinically implemented in November of 2021.Stereotactic Optimized Automated Radiotherapy (SOAR) is a stand-alone application that uses the Eclipse Scripting Application Programming Interface (ESAPI) to design and construct SRS plans based on patient-specific characteristics.SOAR integrates collision prediction, beam angle optimization, and dose optimization into an easily accessible linear workflow.The application creates Volumetric Modulated Arc Therapy (VMAT) plans for delivery on a linear accelerator equipped with a high-definition multi-leaf collimator (HD-MLC) system.
This study seeks to benchmark the novel SOAR automated SRS planning solution against the vendor solution HyperArc.We hypothesized that SOAR would achieve similar plan quality to HyperArc.

Patient and plan characteristics
This study was approved by the Health Research Ethics Board of Alberta -Cancer Committee.A retrospective comparison was performed to determine plan quality of the SOAR automated planning solution compared to the commercial alternative HyperArc.For inclusion in this study, patients were required to have had an SRS or FSRT treatment plan created using the SOAR application.No special exclusion criteria were used, such that patients were not excluded for previous irradiation, targets close to organs at risk, multiple isocenter plans, or for varying dose prescriptions in plans with multiple targets.Gross Tumor Volumes (GTVs) were contoured on gadolinium enhanced T1 MRI images.MRI images were rigidly registered to the planning CT images for treatment planning.Planning CT images were acquired on Philips Brilliance Big Bore scanners using a 1 mm slice thickness.Patients were immobilized with an open-face thermoplastic mask on the Orfit All-in-One board.The CT image volume extended from the top of the Orfit board to the bottom of the chin.Planning Target Volumes (PTVs) were created using a 1 mm margin expansion from the GTV border.The median number of targets treated in this study was three (Range: 1 to 13).Other plan characteristics including the PTV volumes and isocenter off-axis distances can be found in table 1.The prescription dose was prescribed to the 80% isodose line for all plans with a median prescription dose of 22 Gy.All clinical SRS plans were delivered on a Truebeam Edge linear accelerator with HD-MLCs using a 10 MV flattening filter free beam with a max dose rate of 2400 MU/min.

SOAR automated planning
SOAR combines collision prediction, beam angle optimization, and automated plan creation within an easy-to-use application developed using the Eclipse Scripting API. Figure 1.details the sequential steps for automated plan creation and the associated SOAR user interface.Individualized collision prediction using patient CT contours and detailed models of the LINAC gantry, couch, and immobilization structures (figure 1(a)) are the foundation for calculating the available treatment space for gantry trajectories (Mann et al 2019).Manual selection of the targets to be included for the treatment, the OARs to avoid, and the number of isocenters to use for the plan was required before beam angle optimization (figure 1(b)).The choice of isocenter count could also be automated based on the estimated field size of each possible lower dose-volume objectives were added to all PTV and GTV structures using the attached RT prescription to match prescription doses to selected targets.Upper dose-volume objectives were added to selected OARs and the auto-created ring structures.A Normal Tissue Objective (NTO) was also added.Plans were VMAT optimized using the Eclipse Progressive Resolution Optimizer (PRO) v13.6 algorithm.Dosevolume objectives were adjusted until a clinically acceptable plan that met all institutional dose guidelines was achieved.After rigorous quality control, all SOAR plans were delivered for patient treatments.

HyperArc automated planning
HyperArc plans require the Qfix Encompass immobilization system.A duplicate structure set was created from each clinical plan and the Encompass structure was added and aligned with the patient's head.All HyperArc plans used the same prescription doses and fractionation as the clinical plans and a 10MV flattening filter free beam with a max dose rate of 2400 MU/ min.HyperArc plans were limited to a single isocenter, except in cases where multiple fractionations were required.For these cases targets were manually grouped according to fractionation into a single isocenter for each fractionation scheme.Four out of fifty HyperArc plans had two isocenters while the remaining plans used a single isocenter.The number of arcs chosen for each plan was left to the planner's discretion with a maximum of five half arc trajectories.HyperArc automatically limits available arcs if the isocenter is outside the central patient protection zone.
A lower dose GTV objective was added for larger targets (>1 cm 3 ).All HyperArc plans were optimized using the Eclipse Automatic Lower Dose Objective (ALDO) and SRS NTO available with the Eclipse Photon Optimizer v15.6 algorithm.These functions are only accessible with HyperArc planning.If necessary, upper dose constraints were manually added to limit OAR dose.The endpoint for HyperArc dose optimization was acceptable coverage for all targets, and sparing of OARs in each plan, as per institutional guidelines.For targets with competing objectives, such as a target within the brainstem, coverage was allowed to be lower compared to other targets within the plan.Plans were then normalized to match clinical target coverage which was defined as meeting or exceeding the target coverage achieved in the clinical SOAR plans.Normalization was required to be within ±5% of the original plan dose.HyperArc plans were reviewed by a radiation oncologist to ensure clinical acceptability.

Retrospective plan comparison
Our retrospective comparison included SRS plan quality and complexity metrics extracted for both SOAR and HyperArc plans.Plan quality was assessed using radiation dose-volume metrics for the target volumes, the normal brain tissue (Brain-GTV), and other relevant organs at risk within the brain.Both plan cohorts were scored based on compliance with minor and major organ at risk dose-volume constraints according to institutional guidelines, which are outlined in table 2. These guidelines align closely with the HyTEC reports (Grimm et al 2021).For patients treated with multiple fractionations, the normal brain dose was analyzed separately for each plan using the fractionation specific dose-volume tolerance.
Target coverage with the prescription dose was compared using the Paddick Conformity Index (CI) which has a maximum value of 1.0 with ideal coverage.CI values greater than 0.8 are preferred but not always achievable.Dose fall-off was compared using a Gradient Index (GI) which is ideally less than 2 for SRS planning.Plan complexity was assessed using aperture, MLC, gantry speed, and dose rate complexity scores and the modulation factor for each plan.Definitions for all the complexity metrics, conformity index, and gradient index are located in table 3. Dose rate and gantry speed complexity scores are zero for constant dose rate or gantry speeds and increase in value with increasing modulation.A modulation factor less than 4 is preferred clinically.Complexity metrics have been shown to correlate with plan deliverability (Braun, Quirk and Tchistiakova, 2022).Plan metrics were statistically compared using doublesided Wilcoxon signed-rank tests (α = 0.05).A Bonferroni correction was applied due to the use of multiple comparisons, resulting in a lowered significance level of < p .003.Correlations between complexity metrics were investigated using the Pearson correlation coefficient (α = 0.05).Complexity metrics were also compared to historical averages from SRS plans at the Tom Baker Cancer Centre for reference.

Target sphericity
The impact of target shape on radiation conformity to the target was investigated by calculating the sphericity of each target volume.Sphericity is a measure of how closely an object resembles a perfect sphere.Sphericity Y ( ) is calculated by taking the surface area of a perfect sphere with the same volume as an object V , ( ) divided by the actual surface area of the object A .
( ) This metric reaches unity for a perfect sphere and is less than one for any non-spherical object.Surface area for each target was calculated by summing the individual triangle area components that constitute the total structure mesh geometry.Targets were categorized as either low sphericity (Y < 0.9, N = 22) or high sphericity (Y  0.9, N = 157) and differences in target conformity indices between SOAR and Hyper-Arc were compared between both sphericity groups using a Mann-Whitney U test (α = 0.05).

Results
Plan quality Dose-volume results are displayed in figures 2. and 3. using violin plots.These plots combine the summary statistics of a boxplot with a kernel density plot to show variations in the numerical distribution of data.A key identifying the various components of the violin plots is shown in figure 2. The volume of normal brain tissue receiving a threshold dose is categorized based on dose fractionation in figure 2. The single fraction threshold dose is 12 Gy, the three-fraction dose is 20 Gy, and the five-fraction dose is 24 Gy.The median volume of normal brain tissue that received at least 12 Gy in a single fraction (V12Gy) was 7.76 cm 3 for SOAR and 7.47 cm 3 for HyperArc.This difference was statistically significant.Differences in three and five fraction irradiated normal brain volumes were not significant due to the limited sample size.The median differences in max doses to the eyes, lens, optic nerves, and optic chiasm shown visually in figure 3. were statistically significant favoring SOAR.The largest difference for OARs was seen in the median max dose (D0.035cm 3 ) to the Optic Nerves which was 0.96 Gy (range: 0.10 to 7.13 Gy) for SOAR and 1.36 Gy (0.10 to 8.62 Gy) for HyperArc.HyperArc plans showed statistically higher target coverage (V80%) and dose fall-off outside the target (GI).SOAR plans had lower target coverage but better radiation conformity to the target (CI).The sample size, median, minimum, and maximum values for all plan quality metrics can be found in table 4.

Protocol compliance
Plan compliance with institutional dose-volume constraint guidelines was assessed for both HyperArc and clinical SOAR plans.Definitions for minor and major constraints based on structure and dose fractionation  can be found in table 2. The number of constraint violations for all structures is summarized in table 5. Major constraint violations were not included in the count of minor constraint violations for the same structure.The majority of constraint violations occurred in the normal brain structure with both HyperArc and SOAR plans exceeding the minor V12Gy < 10 cm 3 constraint for plans with multiple metastases or large surgical beds.Major constraint violations only occurred in the brainstem and normal brain structures.HyperArc plans showed a reduction in the number of normal brain constraint violations compared to SOAR but had more constraint violations in other structures.

Plan complexity
Plan complexity metrics for both SOAR and HyperArc plans were assessed and compared.A summary of these results can be found in table 6. which shows the median, minimum, and maximum values for each metric.Median values for Aperture, Gantry Speed, and Dose Rate complexity scores were lower in SOAR plans showing reduced complexity.The gantry speed complexity score was the only statistically significant result with a median of 2.42 and max of 7.84 for SOAR compared to 5.14 and 10.33 with HyperArc.Median values for gantry speed and aperture complexity for both planning methods were higher than the historical averages of 1.98 and 19.96 for SRS plans at the Tom Baker Cancer Centre.Median dose rate complexity for both methods was similar to the historical average of 3.29.Median MLC complexity in this study was lower than the historical average of 17.21.The modulation factor was similar between SOAR and HyperArc plans with median values of 3.3 and 3.4 respectively.A strong positive correlation was found between the modulation factor and MLC complexity score for both SOAR (r45 = .74,p < .001)and HyperArc (r45 = .60,p < .001)plans.HyperArc plans also showed a strong negative correlation between dose rate complexity and gantry speed complexity (r45 = −.63,p < .001),however this trend was not seen in the SOAR cohort.

Target sphericity
The impact of target sphericity on conformity was investigated.These results are displayed in figure 4. which shows the percent difference in CI versus target sphericity.The average difference in CI was 2.3% showing overall better conformity for SOAR as already summarized in table 4. The distribution of points with sphericity less than 0.9 had an average difference in CI of 2.45% with positive numbers denoting better SOAR performance.For points greater than 0.9 sphericity the average difference in CI was 2.28%.Differences in CI for low versus high sphericity targets were compared using a Mann-Whitney U test which showed that both distributions were statistically the same.

Discussion
A retrospective study was performed to compare the novel SOAR automated SRS planning solution, recently implemented at the Tom Baker Cancer Centre, to the vendor solution HyperArc.This plan comparison used a large cohort of 50 patients to ascertain differences in plan quality and complexity.SOAR showed better organ-at-risk sparing for the brainstem and cochlea, and statistically significant sparing for the eyes, lens, optic chiasm, and optic nerve structures.This resulted in two minor constraint violations for HyperArc in the lenses and two major constraint violations in the brainstem compared to only one minor lens violation and one major  Plan complexity is correlated with clinical deliverability of the plan (Braun, Quirk and Tchistiakova, 2022).Using an aperture complexity score decision threshold, Younge et al 2016 were able to correctly predict 44% of failed VMAT QA plans with only a 7% false-positive rate (Younge et al 2016).This study compared various complexity metrics for SOAR and HyperArc plans and found that SOAR plans scored lower compared to HyperArc for all complexity metrics.This difference was statistically significant for the gantry speed complexity metric.MLC and aperture complexity could possibly be reduced by utilizing the Aperture Shape Controller setting available in the Eclipse Photon Optimizer v15.6 algorithm which was not used for this study.A further limitation of this study is that HyperArc plans were not delivered on the treatment unit to compare quality assurance results.A study by Wong et al 2022 on the clinical implementation of HyperArc did find lower average film dose passing rates with gamma analysis for HyperArc compared to manual VMAT (Wong et al 2022).Average passing rates for 11 patients at 3%/1 mm gamma criteria were 91.3% for manual VMAT and 88.6% for HyperArc.Popple et al 2021 reported that with 3%/ 2 mm passing criteria, 97% of HyperArc plans had >95% pass rates compared to 94% for manual plans (Popple et al 2021).
An ideal SRS target is spherical in shape, small, far from OARs, and spaced apart to avoid dose bridging.However, this is not always the case.SRS is often used to treat surgical beds or larger metastases which can vary greatly in size and shape.An investigation of radiation conformity based on target shape was included in this study.The plot of target conformity difference versus sphericity in figure 4. appears to show a small benefit for SOAR with less spherical targets having improved target conformity compared to HyperArc.In the 0.9 to 1.0 high sphericity range the conformity difference between HyperArc and SOAR plans shows significant variations in which planning method is achieving better conformity.Under 0.9 sphericity the distribution of points shifts to favor the SOAR planning method.These differences were not found to be statistically significant, however this investigation was limited by the small number of targets included in the low sphericity group (N = 22).The SRS NTO, while effective at creating steep dose distributions, may struggle with maintaining conformity with less spherical targets.With HyperArc planning it was found that the SRS NTO would prioritize limiting bridging dose and compromise dose coverage for targets in a cluster.Additionally, in single isocenter SRS treatments for multiple brain metastases, rotational set-up errors may impact the accuracy of the delivered dose distribution.Small rotational errors introduced during patient setup are exacerbated at further distances from the plan isocenter. .This cohort also included a more diverse population with regards to plan characteristics.Plans treated up to 13 metastases, had targets located in the brainstem, and large surgical beds up to 85.9 cm 3 in size.Both planning methods were tested using a diverse selection of metastatic SRS cases.The SOAR application provided versatility in SRS planning with a wider selection of optimized beam angles to avoid OARS and find optimal collimator angles that minimize island blocking.SOAR also supported multi-iso planning with a visual display to avoid overlapping beams.Limiting the available beam angles with a template-based solution simplifies the beam selection process but limits options in more complicated cases such as avoiding previous sites of irradiation or planning multiple isocenters with overlapping beam angles.Collision prediction with SOAR used patient-specific models of the body and accurate gantry and couch models to individually determine the usable treatment space for each patient as opposed to the coarse patient protection zone strategy used by HyperArc.SOAR did not require a specific immobilization system while HyperArc requires users to purchase the QFix Encompass immobilization system.

Conclusions
SOAR plan quality was compared to the commercial solution HyperArc in a retrospective planning study.SOAR provided greater versatility in SRS planning and better sparing of critical structures within the brain.HyperArc achieved better normal brain sparing and reduced low dose spread but may struggle to achieve appropriate target conformity in difficult cases with non-spherical metastases or targets close to the brainstem.HyperArc may facilitate faster SRS planning but provides limited beam angle options.Both automated planning solutions were shown to be useful in producing reliable high quality SRS plans.SOAR provides a versatile non-commercial automated solution for multi-metastatic SRS treatment planning applicable to any centre using the Eclipse treatment planning system and Truebeam Edge delivery platform.

Figure 1 .
Figure 1.SOAR user interfaces during automated plan creation which are displayed sequentially.(A) Alignment of body and immobilization contours for collision prediction.(B) Parameter selection window for choosing targets, OARs to avoid, and number of isocenters to use.(C) Beam selection window to automatically or manually choose beam angles based on optimization results.(D) Visual display for multi-isocenter planning that shows if arcs overlap.
Figure 2. Violin Plots of the irradiated normal brain volume organized according to fractionation.SOAR is shown in blue and HyperArc in green.A plot key for the violin plots is included on the far right.

Figure 3 .
Figure 3. Various organ-at-risk doses displayed using violin plots.Dose-volume metric details for each OAR can be found in table 4. See figure 2. for a plot key identifying the components of the violin plots.
(Briscoe,  Voroney and Ploquin, 2016).Sagawa et al 2019 examined the effect of rotational setup errors on HyperArc plans and found significant reductions in target coverage with D95% dropping by up to 60% and reductions in plan conformity (Sagawa et al 2019).Direct comparison of SOAR and HyperArc plan times was not possible as HyperArc has not been clinically implemented at the Tom Baker Cancer Centre.Popple et al 2021 found no significant difference in plan times between manual planning and HyperArc for experienced planners with median plan times of 72 and 77min respectively (Popple et al 2021).HyperArc planning was faster than SOAR for the time required to create a treatment plan, taking 4 to 8 min with HyperArc compared to 6 to 12 min for SOAR.This is because SOAR performs trajectory optimization, and collimator angle optimization for the entire treatment space, while HyperArc only performs collimator angle optimization for a maximum of four pre-set table angles.Treatment time was not included in this study, but HyperArc does benefit from automated delivery of imaging and treatment fields.Popple et al 2021 reported a median total treatment time of 10.5 min with HyperArc compared to 11.4 min for manual plans (Popple et al 2021).This study was the first comparison of HyperArc plan quality to an automated beam angle optimized VMAT planning method.The sample size of 50 patients was larger than most HyperArc planning studies (Ohira et al 2018, Slosarek et al 2018, Kadoya et al 2019, Ruggieri et al 2018, 2019, Ueda et al 2019, Vergalasova et al 2019, Fix et al 2020, Wong et al 2022)

Figure 4 .
Figure 4. Percent difference in Conformity Index (CI) between SOAR and HyperArc plans based on target sphericity.Positive values show better conformity for SOAR planning.

followed by beam angle optimization as described in Mann et al 2022 (Mann, Thind and Ploquin, 2022). The optimization process used MLC open field area, OAR overlap with targets, and gantry trajectory range to score beams and rank them relative to one another. Optimization results were displayed in
Eclipse scripting facilitated plan creation, named according to institutional protocol, addition of all the treatment beams with respective gantry, table, and collimator angles, and the addition of VMAT optimization objectives.Dose-limiting ring structures were automatically created as well as group PTV structures for each isocenter.A standard set of VMAT upper and

Table 2 .
Major and Minor SRS and FSRT dose-volume constraint definitions.

Table 3 .
Plan quality and complexity metric descriptions.Overlap of the prescription isodose volume and target volume squared, divided by the prescription isodose volume and target volume Gradient Index 50% prescription isodose volume equivalent sphere diameter divided by the prescription isodose volume equivalent sphere diameter Aperture Complexity Score

table 4 .
See figure 2. for a plot key identifying the components of the violin plots.
3 established by Blonigen et al 2010 (Blonigen et al 2010).HyperArc was able to reduce normal brain constraint violations compared to SOAR with two less minor violations and one less major violation.Normal brain constraint violations with both planning methods occurred in patients with a high number of metastases or large surgical bed targets.HyperArc plans exhibited steep dose fall-off outside the target as shown by the reduced gradient index.This result was consistent with previous studies comparing HyperArc to manual VMAT planning (Ohira et al 2018, Ruggieri et al 2018, Vergalasova et al 2019, Popple et al 2021).

Table 4 .
Plan quality metric results from retrospective plan comparison.

Table 5 .
Count of minor and major dose-volume constraint violations for each structure.

Table 6 .
et al 2021 The new SRS/FSRT technique HyperArc for benign brain lesions: a dosimetric analysis Sci.Rep. 11 1-11 Kadoya N et al 2019 Automated noncoplanar treatment planning strategy in stereotactic radiosurgery of multiple cranial metastases: HyperArc and CyberKnife dose distributions Medical Dosimimetry 44 394-400 Mann T D et al 2019 Development and clinical implementation of eclipse scripting-based automated patient-specific collision avoidance software Journal of Applied Clinical Medical Physics 20 12-9 Mann T D, Thind K S and Ploquin N P 2022 Fast stereotactic radiosurgery planning using patient-specific beam angle optimization and automation Physics and Imaging in Radiation Oncology 21 90-5 Ohira S et al 2018 HyperArc VMAT planning for single and multiple brain metastases stereotactic radiosurgery: A new treatment planning approach Radiation Oncology 13 1-9 Popple R A et al 2021 Transition from manual to automated planning and delivery of volumetric modulated arc therapy Plan complexity metric results.statistically significant results are bolded.stereotactic radiosurgery: clinical, dosimetric, and quality assurance results Practical Radiation Oncology 11 e163-71 Ruggieri R et al 2018 Linac-based VMAT radiosurgery for multiple brain lesions: Comparison between a conventional multiisocenter approach and a new dedicated mono-isocenter technique Radiation Oncology 13 1-9 Ruggieri R et al 2019 Linac-based radiosurgery for multiple brain metastases: Comparison between two mono-isocenter techniques with multiple non-coplanar arcs Radiotherapy and Oncology 132 70-8 Sagawa T et al 2019 Dosimetric effect of rotational setup errors in stereotactic radiosurgery with HyperArc for single and multiple brain metastases Journal of Applied Clinical Medical Physics 20 84-91 Slosarek K et al 2018 In silico assessment of the dosimetric quality of a novel, automated radiation treatment planning strategy for linac-based radiosurgery of multiple brain metastases and a comparison with robotic methods Radiation Oncology 13 1-10 Ueda Y et al 2019 Dosimetric performance of two linear acceleratorbased radiosurgery systems to treat single and multiplebrain metastases The British Journal of Radiology 92 1-9 Vergalasova I et al 2019 Multi-institutional dosimetric evaluation of modern day stereotactic radiosurgery (SRS) treatment options for multiple brain metastases Frontiers in Oncology 9 1-12 Wong F H C et al 2022 Clinical implementation of HyperArc Physical and Engineering Sciences in Medicine 45 577-87 Yamamoto M et al 2014 Stereotactic radiosurgery for patients with multiple brain metastases (JLGK0901): A multi-institutional prospective observational study The Lancet Oncology 15 387-95 Younge K C et al 2016 Predicting deliverability of volumetricmodulated arc therapy (VMAT) plans using aperture complexity analysis Journal of Applied Clinical Medical Physics 17 124-31 10 Biomed.Phys.Eng.Express 10 (2024) 025037 T Mann et al a