Tuning of AcurosXB source size setting for small intracranial targets

Abstract This study details a method to evaluate the source size selection for small field intracranial stereotactic radiosurgery (SRS) deliveries in Eclipse treatment planning system (TPS) for AcurosXB dose calculation algorithm. Our method uses end‐to‐end dosimetric data to evaluate a total of five source size selections (0.50 mm, 0.75 mm, 1.00 mm, 1.25 mm, and 1.50 mm). The dosimetric leaf gap (DLG) was varied in this analysis (three DLG values were tested for each scenario). We also tested two MLC leaf designs (standard and high‐definition MLC) and two delivery types for intracranial SRS (volumetric modulated arc therapy [VMAT] and dynamic conformal arc [DCA]). Thus, a total of 10 VMAT plans and 10 DCA plans were tested for each machine type (TrueBeam [standard MLC] and Edge [high‐definition MLC]). Each plan was mapped to a solid water phantom and dose was calculated with each iteration of source size and DLG value (15 total dose calculations for each plan). To measure the dose, Gafchromic film was placed in the coronal plane of the solid water phantom at isocenter. The phantom was localized via on‐board CBCT and the plans were delivered at planned gantry, collimator, and couch angles. The planned and measured film dose was compared using Gamma (3.0%, 0.3 mm) criteria. The vendor‐recommended 1.00 mm source size was suitable for TrueBeam planning (both VMAT and DCA planning) and Edge DCA planning. However, for Edge VMAT planning, the 0.50 mm source size yielded the highest passing rates. The difference in dose calculation among the source size variations manifested primarily in two regions of the dose calculation: (1) the shoulder of the high‐dose region, and (2) for small targets (volume ≤ 0.30 cc), in the central portion of the high‐dose region. Selection of a larger than optimal source size can result in increased blurring of the shoulder for all target volume sizes tested, and can result in central axis dose discrepancies in excess of 10% for target volumes sizes ≤ 0.30 cc. Our results indicate a need for evaluation of the source size when AcurosXB is used to model intracranial SRS delivery, and our methods represent a feasible process for many clinics to perform tuning of the AcurosXB source size parameter.


| INTRODUCTION
Stereotactic radiosurgery (SRS) has become a valuable treatment modality to treat lesions within the brain 1 and spine. 2 In particular, SRS provides a non-invasive treatment approach for unresectable tumors (such as those in the eloquent cortex or otherwise deepseated tumors) or for patients who are otherwise not candidates for surgery. 3 Though SRS was first performed using a specialized device, now commercially available as the GammaKnife, technological advances have allowed for stereotactic therapies using the linear accelerator.
Several noteworthy advances have allowed for improved precision in linear accelerator-based SRS: (1) the advent of treatment room stereotactic imaging systems, [4][5][6][7] including stereoscopic planar imaging and cone-beam CT imaging, (2) improvements in patient support devices, including 6 degree-of-freedom capabilities in the treatment couch 8 and improvements in couch movement precision, (3) increasing availability of high-intensity photon modes, [9][10][11] such as flattening-filter free photon modes with dose rates up to 2400 MU/min, (4) high-definition multi-leaf collimator (MLC) systems with leaf widths as narrow as 2.5 mm, 12,13 and (5) optical monitoring systems to track patient motion throughout the treatment course. 14,15 Of course, the advances in the preceding paragraph all focus on the treatment delivery, while an accurate end-to-end treatment delivery relies on the marriage of the dose modeling within the treatment planning system and the physical dose delivery within the treatment room. Along these lines, the modeling of small-field dose delivery has garnered much interest. The accurate measurement and modeling of small-field dose delivery (i.e., field sizes < 3 9 3 cm 2 in water-equivalent media) has many challenges, including the effects of the finite size of the radiation source, loss of charged particle equilibrium (CPE), and sensitivity to small changes in field size for perturbation factors of ion chambers used for measurement. 16 The AcurosXB algorithm gives a discretized solution to the linear Boltzmann transport equation, [17][18][19] which provides improvements in regions with loss of CPE, such as heterogeneity interfaces. 20,21 However, the proper selection of source size within the AcurosXB algorithm is still essential for the accurate modeling of small field deliveries. The purpose of this study was two-fold: (1) to present a clinically achievable method to evaluate the source size for smallfield dose calculation, and (2) to use the method to evaluate the ideal source size setting within AcurosXB for flattening-filter energy modes for two delivery platforms (Varian Edge and TrueBeam machines), MLC leaf models (Millennium120 HD-MLC and standard Mil-lennium120 MLC), and delivery techniques for intra-cranial SRS planning (DCA and VMAT). All planning and dose calculations were performed using Varian Eclipse (v. 13) TPS and AcurosXB dose calculation algorithm (v. 13; 1.0 mm isotropic dose calculation grid size and dose to medium calculation setting). The dose to medium setting was used in accordance with recently published recommendations from the NRG physics group. 22 The current study was designed to evaluate the source size setting for small intracranial targets, with target volumes as small as 0.03 cc (which corresponds to approximately 4 mm diameter); in addition, several other studies have used smaller dose grid size (i.e., less than 1.5 mm) as the standard for dose calculation comparisons when evaluating small-field dosimetry. [23][24][25][26] With this in mind, this study used 1.0 mm dose grid size for all dose calculations.
All planning was done with 6x-FFF beam energy with nominal dose rate set to the maximum setting (1400 MU/min). The patient treatment plan was generated with beam model parameters following the vendor recommendations for source size (spot size setting of 1.00 mm in X-and Y-directions) and our current clinical values for MLC parameters (i.e., dosimetric leaf gap (DLG) and MLC leaf transmission value). 7 The details of the relevant treatment planning data (including target volume size and location) are shown in Table 1 for   Edge linac and Table 2 for TrueBeam linac.
To study the influence of source size on the calculated dose for cranial SRS deliveries, a total of 5 AcurosXB beam models were created for each machine type. The user can tune the source size in the beam configuration module through varying the effective target spot size value, which is entered by the user separately for X-and Ydirections. 19 All beam models used the same input measured data (percent depth-dose, cross-line profiles, and output factors), with the source size varied from 0.50 mm to 1.50 mm in 0.25 mm increments for each machine type. Each beam model was then calculated separately with its unique source size value. In addition to the source size parameter, the DLG was also varied in the analysis. In Eclipse, the DLG represents the TPS method for modeling of the rounded MLC leaf end. 19 For small MLC-defined fields, the leaf end modeling and the potentially partial viewing of the finite size of the radiation source along the central axis are inherently coupled. The DLG parameter is not included within the Beam Configuration workspace in ARIA v. 13; rather, it is included in the machine properties of the RT Administration workspace. Nonetheless, the DLG parameter was also varied in our analysis: three DLG values were included in the modeling and calculation analysis for each machine type. GARDNER ET AL.

| 171
All treatment plans were mapped to a water-equivalent slab phantom, with total phantom dimensions of 15 cm 9 30 cm 9 30 cm (Gammex Inc., Middleton, WI, USA). The isocenter was placed in the center of the phantom, corresponding to 7.5 cm depth. Each treatment plan was calculated for all combinations of source size beam model and DLG value. Thus, for each treatment plan, a total of 15 dose calculations were performed to sample the various source size and DLG values for dose calculation. All dose calculations were performed with the same monitor unit values determined during the original plan optimization. After dose calculation was completed, planar dose planes were exported for analysis (512 9 512 matrix resolution, 5 cm 9 5 cm matrix size).

2.B. | Gafchromic film measurements and calibration
Film measurements using Gafchromic EBT3 film (Film Size: 20.3 9 25.4 cm 2 ; Ashland Inc., Covington, KY, USA) were used to evaluate the dose calculation accuracy in this study. Gafchromic film was selected due to several attractive detector properties: extremely high spatial resolution, large planar detection area, minimal directional dependence, and low energy dependence. In addition, Gafchromic film and associated dosimetric analysis tools are widely available to the radiation oncology community, making our methods described here feasible for many clinics. The films were handled according to the recommendations of AAPM Task Group 55. 27 The phantom localization and treatment procedure followed our clinical process for intracranial SRS treatment delivery. Specifically, the phantom and film plane (coronal plane at mid-phantom -7.5 cm depth) were localized using CBCT imaging prior to dose delivery (125 kVp, Full-fan filter, 1 mm slice thickness), and all plans were delivered at planned gantry, collimator, and couch angles. The average delay time between irradiation and film scanning was approximately 24 hr. Films were scanned in an Epson Expression 10000XL flatbed scanner (Seiko Epson Corp, Nagano, Japan). All films were scanned at the center of the scanner bed with resolution settings of 150 dot per inch and 48 bit RGB mode (16 bits per color channel). A four-way flip method was used to average out any intrinsic light source non-uniformity of the scanner. Dosimetric analysis was done via green channel due to its superior sensitivity at the dose levels larger than 10 Gy. 28 The film calibration and dosimetric analysis was performed using in-house software. Calibration films were irradiated in a nine square dose pattern (area of 2 9 2 cm 2 per square). The in-house calibration routine matches the film optical densities within each square to the TPS calculated dose for the same beam geometry. Then, a

2.C. | Dose distribution analysis
The film measurements were compared to the calculated dose planes using Gamma analysis. 29 Typical Gamma analysis for patient-specific IMRT QA may use distance-to-agreement criteria of 2-3 mm. However, the measurement scale (percentage of measurement points with passing Gamma values) is often saturated if typical Gamma criteria are used. To determine the appropriate Gamma analysis criteria for this study, the Gamma analysis passing rate results for two representative cases were logged for a variety of dose difference and distance-to-agreement criteria and compared to qualitative visual dose profile analysis. The best agreement between passing rate result and visual profile analysis was found for the following Gamma criteria: 3% dose difference and 0.3 mm distance-to-agreement. It should be noted that these criteria are likely too strict for planning with conventional target sizes. But, for very small targets such as those found in intracranial SRS planning, a distance-to-agreement criteria of 1 mm is quite large relative to the lesion radius (for example, the radius of a 0.5 cc spherical lesion is approximately 5 mm). See Fig. 1 for comparison of the Gamma criteria for one of the representative cases. The value for dose threshold was set to 20% of the maximum film plane dose, which corresponds to roughly 25% of the prescription dose for these cases.

2.D. | Statistics
Gamma analysis passing rate results for each source size setting were compared to passing rate results for vendor-recommended source size setting using Student's t-test, assuming two-tailed distribution with P < 0.05 significant.

3.A. | Film dosimetry resultsgamma analysis
The Gamma analysis passing rate results for Edge linac are shown in Comparison of Gamma analysis criteria for one representative case from the study. The red regions in each Gamma map represent failing pixels for the relevant Gamma criteria used. a, Planned dose (1.50 mm source size) plane with line profile geometry (green horizontal line). b, Green channel film dose plane. c, Gamma map for 1%,1 mm criteria. d, Gamma map for 2%, 1 mm criteria. e, Gamma map for 3%, 1 mm criteria. f, Gamma map for 3%, 0.5 mm criteria. g, Gamma map for 3%, 0.3 mm criteria. h, Line profile comparing the AcurosXB planned dose (1.50 mm source size) and green channel film dose. Note the disagreement between planned and measured dose in the shoulder of the high-dose region. Gamma analysis using 1 mm dose-to-agreement criteria is insensitive to such discrepancy in the dose distribution, while Gamma analysis with tighter distance-to-agreement criteria (e.g., 0.3 or 0.5 mm) shows failing points in the shoulder of the high-dose region that match observed dose distribution discrepancies. The Gamma analysis criteria used for this study: 3%, 0.3 mm.
whiskers indicate the maximum and minimum passing rates for each source size setting and DLG value. The first, second, and third quartile values for each combination of settings are also displayed on the plot.  high-intensity flattening filter free energy modes, [9][10][11] and surface imaging systems for tracking. 14,15 With improvements in the localization and delivery systems, there remains a definite need for accurate modeling of the small field dose delivery within the treatment planning system. The purpose of this study was two-fold: (1) to present a clinically achievable method to evaluate the source size for small-field dose calculation, and (2) 24 Their study included testing the source size (focal spot size) for values of 0 mm, 1 mm, and 2 mm. They found agreement in output factor prediction between AAA and AcurosXB to be within 1% for field sizes ≥   The use of gamma analysis for IMRT QA has been the subject of much scrutiny. [33][34][35][36][37][38] In particular, gamma analysis using traditional criteria for distance-to-agreement (on the order of 2-3 mm) and dose difference (on the order of 2-3%) may not be sensitive to clinically meaningful dose errors when per-beam IMRT analysis is used. We note several considerations regarding the use of gamma analysis in this study. First, gamma analysis provides a binary result for each pixel (i.e., the pixel either passes or fails the test), and the gamma analysis does not discriminate between delivered dose that is higher or lower than the planned dose. For this reason, commissioning of small-field deliveries should not rely on gamma passing rates alone; rather, the gamma map and line profile analysis should also be used to give a better understanding of the agreement between planned and measured doses. In this study, we present gamma analysis passing rates and line profile analyses of representative cases. Second, all dose distributions analyzed in this study are composite dose distributions. The dose distributions for each arc are summed in the phantom just as they would in a clinical patient. We believe this avoids one major issue with typical IMRT QA, which is the lack of correlation between per-beam planar measurements and clinically meaningful dose errors. Third, our institutional film analysis program allowed for the use of relatively small distance-to-agreement criteria (on the order of 0.3 mm) for these cases; similar distance-to-agreement criteria (0.5 mm) was previously used to validate GammaKnife dosimetry. 39 The details of our institutional practice for SRS/SBRT QA using Gafchromic film have been published 40   calculations, film dosimetry, or other high-resolution dosimetric data. 17,24,32 The difference in the film dosimetry results in the current study for VMAT and DCA planning groups (particularly for Edge linac with high-definition MLC) underscore the need to extend the dose calculation model analysis to include intensity-modulated deliveries.
The modeling of small-field deliveries within Eclipse is a combination of the field-specific output factor (determined from the collimator back scatter factor (CBSF) table in Eclipse), modeling of the MLC leaf end (primarily determined from the DLG value in Eclipse), and the modeling of the source. This study analysis included evaluation of the latter two parameters, but did not fully consider the effects of the CBSF table. However, all beam models were generated using output factor down to jaw sizes of 1 9 1 cm 2 , with small field data measured using a stereotactic field diode. It is important to note that all plans analyzed in this study utilized jaw settings larger than 1 9 1 cm 2 ; the smallest jaw setting for this study was 1.6 cm 9 1.4 cm (X by Y). Additionally, MLC-defined small field delivery was validated for field sizes down to 5 mm 9 5 mm using multiple detectors. During commissioning, all small field data was measured multiple times and cross-compared to several detectors for validation, and the calculated output factor data compared favor-

| CONCLUSION
This study highlights the need for tuning of the radiation target source size for the AcurosXB dose calculation algorithm in the context of intracranial SRS dose delivery using DCA and VMAT. In particular, we note the differences in optimal source size values for high-definition (2.5 mm leaf width) and standard (5 mm leaf width) MLC with flattening-filter free delivery. Improper selection of the source size can affect the accuracy of the shoulder of the high shoulder for a wide range of intracranial target sizes, and can also have a drastic effect on the magnitude of the central high-dose region for very small targets (target volume ≤ 0.30 cc).

ACKNOWLEDG MENT
This research was supported, in part, by a grant from Varian Medical Systems (Palo Alto, CA, USA).

CONFLI CT OF INTEREST
This work has been supported in part by a grant from Varian Medical Systems, Palo Alto, CA.