Comparative Analysis of Commercial Independent Dose Calculation Systems for Gamma Knife Stereotactic Radiosurgery

Abstract

Purpose:

Independent dose calculation is recommended for stereotactic radiosurgery (SRS) quality assurance. This study evaluates the accuracy of three commercial secondary dose calculation software programs-MUCheck, RadCalc, and ClearCalc-against the Elekta Leksell Gamma Plan (LGP) treatment planning system (TPS) for Gamma Knife SRS.

Materials and Methods:

Dose calculations for 133 patients (239 intracranial targets) planned with the LGP TPS using the TMR10 algorithm were retrospectively analyzed with MUCheck, RadCalc, and ClearCalc. Point dose calculations were compared between each software and LGP. Statistical analysis included analysis of variance (ANOVA) and paired t-tests, with clinical tolerance assessed against the American Association of Physicists in Medicine (AAPM) Task Group (TG)-178 recommendations.

Results:

Mean differences were 0.34% ± 1.10% (MUCheck), 0.53% ±0.53% (RadCalc), and 0.63% ± 0.16% (ClearCalc). All showed statistically significant differences from LGP but remained within clinical tolerance. Clinical tolerance analysis showed 97.9%, 100%, and 100% of calculations within ±3% for MUCheck, RadCalc, and ClearCalc, respectively. ANOVA revealed significant differences among platforms (F = 10.46, P < 0.001), although all deviations remained clinically acceptable.

Conclusion:

All three software platforms provide clinically acceptable independent dose verification for Gamma Knife SRS, with deviations well below AAPM TG-178 recommended tolerance levels. Selection of the appropriate secondary dose calculation platform should be based on institutional workflow integration and specific clinical requirements rather than statistical differences alone.

INTRODUCTION

Stereotactic radiosurgery (SRS) is a specialized form of radiation therapy that delivers focused high-dose radiation precisely to small intracranial targets in a single or limited number of fractions. This technique relies on steep dose gradients to achieve therapeutic doses within the target while minimizing dose to surrounding healthy brain tissue. Leksell Gamma Knife (LGK) (Elekta A.B., Stockholm, Sweden) is a teletherapy machine that specializes in SRS of the brain. The current platform and techniques used to treat SRS patients were originally developed by neurosurgeon Lars Leksell in the early 1950s.[1,2] The latest LGK models are equipped with 192 sealed Co-60 sources arranged in eight motorized sectors, each sector containing 24 sources. These sectors move over a drilled tungsten collimation ring system, providing 4-, 8-, and 16 mm diameter circular field sizes at the radiological focal point of the machine.[3] The latest models, Icon™ and Esprit™, also include a cone-beam computed tomography (CT) system for frameless radiosurgery treatments and a high-definition motion management camera for intrafraction patient monitoring.[4,5]

Leksell Gamma Plan (LGP) serves as the dedicated treatment planning system (TPS) for LGK. The LGP TPS allows users to shape the dose by superimposing focused radiation of different collimator sizes and sector blocking patterns within the target volume.[6] The latest version (version 11.3.2) of the TPS provides an inverse planning tool, Gamma Knife Lightning, to improve the plan quality and reduce intra-user variability in treatment planning.[7,8,9] Dose distribution throughout the treatment volume is calculated using either a TMR10 algorithm or a convolution algorithm. The TMR10 algorithm is based on ray tracing and precalculated off-axis ratios, whereas the convolution algorithm is based on collapsed cone convolution calculation methodology.[10,11,12,13]

The importance of dosimetric accuracy in SRS has driven the development of independent secondary calculation systems.[14,15] The American Association of Physicists in Medicine (AAPM) Task Group (TG) 178 recommends a tolerance level of 3% and an action level of 5% for an independent check of dose calculations for LGK radiosurgery.[16] These recommendations are based on limiting systematic errors in dose calculation that could lead to serious clinical consequences when treating small targets adjacent to critical structures.

Several institutions have developed in-house software programs that verify treatment parameters as an independent dose verification for LGK.[17,18,19,20,21,22] In recent years, commercial software solutions such as MUCheck (Oncology Data Systems Inc, Oklahoma City, USA), RadCalc (LAP GmbH, Luneburg, Germany), and ClearCalc (Radformation, New York, USA), have emerged to fulfill this requirement. Previous studies have demonstrated the feasibility and accuracy of commercial secondary dose calculation systems for LGK treatments. Xu et al. reported 2-year clinical experience with MUCheck, demonstrating acceptable agreement with LGP calculations while identifying specific limitations related to gamma angle variations and shallow target locations.[23] Prusator et al. evaluated RadCalc software, showing good agreement with the TPS over a wide range of target and patient setup geometries.[24]

While individual software platforms have been evaluated independently, comprehensive head-to-head comparisons remain limited. This study addresses this gap by comparing three commercially available software platforms for LGK radiosurgery. The objectives of the study were (1) to conduct a comprehensive comparison of the dosimetric accuracy and statistical agreement of the three commercial software programs, MUCheck, RadCalc, and ClearCalc, for independent dose verification of LGP treatment plans and (2) to provide evidence-based guidance for clinical physicists and institutions considering the implementation of a secondary dose verification system for LGK radiosurgery. This analysis aims to inform clinical decision-making by evaluating not only the statistical accuracy of each software program but also their practical implications for routine clinical use.

MATERIALS AND METHODS

Study design and patient selection

This retrospective analysis evaluated dose calculations for 133 patients (67 frame and 66 frameless) encompassing 239 targets (72 targets from the frame cohort and 167 targets from the frameless cohort) treated with the LGK Icon™ system at our institution. All treatment plans were generated using the LGP TPS (version 11.3.2). The study was conducted under institutional review board approval, with a waiver of informed consent for retrospective data analysis. Treatment plans (frame and frameless) included in this analysis represent typical clinical practice at our institution. Target volumes ranged from 0.003 cc to 33.49 cc. Prescription doses ranged from 6 Gy to 80 Gy delivered in a single or fractionated treatment (3–5 fractions). Plans utilized various combinations of 4-, 8-, and 16 mm collimators or blocked sectors.

Dose calculation methodology

Leksell Gamma Plan (Elekta A.B., Stockholm, Sweden)

The LGP TPS for LGK employs the TMR10 algorithm to compute dose distributions and shot times with high precision.[6] This algorithm relies on fundamental physics principles: inverse square law, exponential attenuation in water, output factors, and dose profiles measured for 4-, 8-, and 16 mm collimators. Output factors, calibrated against reference dosimetry, account for Co-60 source strength and collimator-specific beam output, ensuring accurate dose delivery. Skull geometry, derived from stereotactic frame measurements or CT/magnetic resonance imaging (MRI) images, enables the calculation of patient-specific attenuation and depth. The TMR10 model assumes water-equivalent tissue, simplifying calculations but potentially underestimating heterogeneity effects. Although a convolution-based algorithm can model tissue inhomogeneities, it was excluded from this study due to limited clinical adoption. An independent dose calculation software needs to approximate the LGK treatment geometry, account for patient anatomy, and plan complexity to complete a water-based dose calculation for comparison with the LGP TPS TMR10 algorithm calculation.

MUCheck (Oncology Data Systems, Oklahoma City, USA)

MUCheck’s GammaKnifeCheck module (version 9.3.0) utilizes a point-dose calculation method that is the same as the classic TMR10 dose calculation algorithm used by the LGP system.[25] The software determines the point dose at a given reference point by reading the shot information from LGP or by manual entry, and sums the contributions from all shots and source sectors. The software models the hemispherical arrangement of the 192 Co-60 sources across eight sectors and uses ray tracing to calculate individual beam contributions based on depth, off-axis distance, collimator-specific output factors, and inverse square law corrections. Similar to the LGP TPS, the skull geometry for a patient is required for dose calculations. There are two methods for attaining the skull geometry: simulated skull provided by the skull scaling technique, which estimates the skull geometry based on stereotactic radial measurements and image-based with either CT or magnetic resonance images.

RadCalc (LAP GmbH, Luneburg, Germany)

RadCalc’s Gamma Knife module (version 7.3.1) performs a point-dose calculation at the target using a virtual machine model based on Elekta’s TMR10 algorithm.[26] It uses proprietary beam data supplied by Elekta and known source arrangements to independently calculate the dose. Once the plan has been imported via direct LGP database connection, RadCalc retrieves the parameters for each shot (collimator size, shot location, sector patterns, etc.) and sums the contribution of all active Co-60 beams for each shot. External contour reconstruction is an independent process within the software based on either image thresholding or scalar measurements from the placement of the stereotactic frame. The external contour uses ray tracing to determine the depth and off-axis distances, which are used to interpolate dose from the Elekta-provided data tables. This approach provides a quick secondary check of the dose at the prescription point (usually the target center) rather than a full three-dimensional dose distribution. Because the data for the virtual machine model is supplied from Elekta directly, there is no user commissioning data that is input into the secondary dose calculation engine.

ClearCalc (Radformation, New York, USA)

ClearCalc (version 2.5.10) performs TMR10-based analytical calculations using a beamlet methodology that individually models dose contributions from each of the 192 Co-60 sources.[27] The software imports patient-specific skull geometry directly from the LGP database as a segmented triangulated mesh. For each shot, individual beamlets are generated at source positions, then rotated and translated in space according to shot coordinates, collimator size, sector blocking patterns, and gamma angle. The calculation establishes vectors from source to isocenter, raytraces through the imported skull geometry to determine source-to-skull distance and depth to target, applies TMR10 formalism with exponential attenuation and hard-coded manufacturer output factors, and sums all beamlet contributions. A key feature is explicit geometric correction for non-standard gamma angles through beamlet transformation.

Secondary dose verification workflow

Figure 1 shows the independent secondary dose verification workflow for the three software platforms. Following treatment plan approval by the treating physician in the LGP TPS using the TMR10 algorithm, each secondary software platform queries the LGP database, retrieving plan-specific parameters including patient demographics, prescription dose, shot coordinates, collimator settings, sector blocking configurations, and skull geometry. Each software platform computes the point dose using its respective algorithm, as described in the Dose Calculation Methodology subsection. The percent difference between the independent software and the LGP reference dose is evaluated against the AAPM TG-178 tolerance level of 3% and action level of 5%. A verification report is generated in PDF format, imported into the record and verify (R and V) system and approved by the physicist before treatment delivery. Cases exceeding the action level are reviewed by the physicist and any discrepancy is resolved before treatment proceeds.

Figure 1.

Schematic workflow for independent secondary dose verification

Figure 2 shows a representative LGP treatment plan for a single-fraction SRS case, illustrating the planning data used as input for independent secondary dose verification. The prescription dose, shot parameters, collimator sizes, and sector blocking patterns in the treatment plan report represent the specific data elements retrieved by each secondary software platform directly from the LGP database, as described in the Secondary Dose Verification Workflow above. Fractionated GK cases follow the same planning structure, with the prescription dose divided across multiple treatment sessions, typically three to five fractions.

Figure 2.

Representative Leksell Gamma Plan (LGP) treatment plan for a single-fraction stereotactic radiosurgery case. (a) Dose distribution with isodose lines overlaid on magnetic resonance imaging in axial, sagittal, and coronal planes; prescription isodose line is shown in yellow (b) treatment plan report from LGP treatment planning system showing Leksell Gamma Knife machine data, fixation configuration and treatment data including prescription dose, shot parameters, collimator sizes, and sector blocking configurations

Statistical analysis

Dosimetric accuracy was evaluated by comparing each software’s calculated dose at the prescription point to the corresponding LGP TPS reference dose calculation. Percent difference was calculated with Eq. 1

Descriptive statistics including mean, median, standard deviation, and range were calculated for each software platform to characterize the distribution of dose differences.

Statistical analysis included a one-way analysis of variance (ANOVA) test to determine statistical differences between the three independent software evaluated in this study. If a significant difference is found, it suggests that at least one of the software programs produces systematically different dose calculations compared to the others. Paired t-tests were also used to determine differences between each secondary dose calculation software and the LGP TPS. A significant result in this test indicates that a software program consistently calculates doses differently than LGP TPS, although clinical relevance must still be considered. Pairwise t-tests were performed between software platforms (MUCheck vs. RadCalc, MUCheck vs. ClearCalc, RadCalc vs. ClearCalc) to determine whether the platforms differed significantly from each other in their calculation deviations. All statistical analyses were performed using Microsoft Excel (Microsoft Corporation) with the Data Analysis ToolPak. Statistical significance was defined as P < 0.05 for all tests.

Clinical tolerance assessment

Clinical tolerance analysis was performed to evaluate the practical applicability of each software platform for routine clinical use. The proportion of calculations agreeing within clinically established tolerance levels (±1%, ±2%, ±3%, and ± 5%) was determined for each platform. These tolerance levels were selected based on AAPM TG-178 recommendations, which specify a ± 3% tolerance level and ± 5% action level for independent dose calculations. The ± 1% and ± 2% levels were included to assess whether tighter institutional tolerances might be achievable for routine quality assurance.

Gamma angle assessment

The gamma angle describes the vertical orientation of the patient’s head. A standard gamma angle of 90° corresponds to a neutral chin position. Non-standard gamma angles, typically used for target optimization or critical structure avoidance, include 70° for chin extension and 110° for chin tuck. In this study, all the frame cases were planned using the standard gamma angle of 90°. To evaluate the performance of each secondary software at the non-standard gamma angles, a subset of patients (five patients) from the frame cohort were replanned using the non-standard gamma angles (70° and 110°).

RESULTS

Mean differences between the secondary dose calculation software and the LGP TPS were 0.34% ±1.10% for MUCheck, 0.53% ±0.53% for RadCalc, and 0.63% ±0.16% for ClearCalc. Table 1 presents comprehensive statistical analysis results for each secondary dose calculation software, including descriptive statistics and paired t-test results comparing each software to the LGP TPS reference. ANOVA revealed significant differences (P < 0.05) among the three platforms (F = 10.46, P < 0.001), and paired t-tests confirmed that all three software programs showed statistically significant deviations from LGP TPS [Table 1]. However, all differences were below the ± 3% threshold recommended by AAPM TG-178.

Table 1.

Statistical comparison of three secondary dose calculation software platforms against Leksell Gamma Plan for 239 intracranial targets

Software	Mean±SD (%)	Median (%)	Range (%)	95% CI (lower)	95% CI (upper)	t	P	Significance
MUCheck	0.34±1.10	0.27	2.58 to 4.33	0.196	0.476	4.72	<0.001	Highly significant
RadCalc	0.53±0.53	0.50	−0.64 to 2.66	0.467	0.603	15.45	<0.001	Highly significant
ClearCalc	0.63±0.16	0.64	0.03 to 0.99	0.608	0.649	59.66	<0.001	Highly significant

ANOVA: F=10.46, P<0.001. SD: Standard deviation, CI: Confidence interval

Pairwise comparisons using paired t-tests revealed that all three software platforms differed significantly from each other (MUCheck vs. RadCalc: t = −2.51, P = 0.013; MUCheck vs. ClearCalc: t = −4.06, P < 0.001; RadCalc vs. ClearCalc: t = −2.59, P = 0.010). Despite these statistical differences, the clinical implications are minimal as all platforms maintained mean differences below ± 1%.

Box-and-whisker plot analysis, shown in Figure 3, revealed distinct distribution patterns for each software. RadCalc demonstrated a symmetric distribution around the mean with minimal outliers, whereas MUCheck showed a broader distribution with a few outliers beyond ± 3%. ClearCalc exhibited the narrowest distribution among the three software platforms.

Figure 3.

Box-and-whisker plot showing percent dose differences between the Leksell Gamma Plan treatment planning system and three secondary dose verification systems: MUCheck, RadCalc, and ClearCalc. Dashed lines indicate ± 3% clinical tolerance recommended by American Association of Physicists in Medicine Task Group-178. SD: Standard deviation, TPS: Treatment planning system

Clinical tolerance assessment

Assessment of clinical tolerance [Table 2] revealed excellent performance across all three software platforms with all programs exceeding 95% compliance with the ±3% tolerance recommended by AAPM TG-178. ClearCalc demonstrated perfect compliance, with 100% of calculations falling within ±1%, ±2%, and ±3% tolerance levels. RadCalc achieved 100% compliance at ±3% tolerance level and 99.2% at ±2%. Under more stringent ±1% criteria, compliance was 83.7%. MUCheck showed more variable performance, with 69.0% compliance at ±1%, 90.0% at ±2%, and 97.9% at ±3%. Five cases exceeded the ±3% tolerance threshold. Detailed analysis of these outlier cases revealed that the targets were located near the skull surface. All three platforms achieved 100% compliance at the ±5% action level.

Table 2.

Percentage of calculations within specified tolerance levels for each software platform

Tolerance Level (%)	MUCheck (%)	RadCalc (%)	ClearCalc (%)
±1	69.0	83.7	100.0
±2	90.0	99.2	100.0
±3	97.9	100.0	100.0
±5	100.0	100.0	100.0

Histogram analysis [Figure 4] provided additional visualization of the frequency distributions for each software platform. MUCheck demonstrated a near-normal distribution centered at 0.34%, RadCalc showed a symmetric distribution centered at 0.53%, and ClearCalc exhibited a narrow distribution centered at 0.63%.

Figure 4.

Frequency distribution of percent dose differences for MUCheck, RadCalc and ClearCalc. Vertical dashed lines indicate ±3% tolerance levels

Gamma angle assessment

Analysis of the frame cases planned at the non-standard gamma angles showed that MUCheck was sensitive to gamma angles compared to the other two software platforms. The difference at 70° gamma angle for MUCheck was −2.58% ±5.02%, whereas it was 0.53% ±0.51% for RadCalc and 0.67% ±0.12% for ClearCalc. For cases planned at 110° gamma angle, the differences were −0.83% ±6.02% for MUCheck, 0.27% ±0.60% for RadCalc and 0.63% ±0.11% for ClearCalc.

DISCUSSION

This comprehensive evaluation of three commercial secondary dose calculation software programs for LGK radiosurgery provides valuable insights into the current state of independent dose verification systems. The observed mean deviations of 0.34% ±1.10% (MUCheck), 0.53% ±0.53% (RadCalc), and 0.63% ±0.16% (ClearCalc) show excellent agreement with the reference LGP TPS calculations. RadCalc demonstrated the smallest mean deviation, suggesting the most consistent performance across the diverse range of treatment plans analyzed. This consistency likely stems from its use of vendor-supplied beam data and virtual machine modeling, which closely mirrors the proprietary algorithm used in the LGP TPS. MUCheck, while showing the largest standard deviation, demonstrated excellent median agreement with LGP calculations. ClearCalc exhibited the most consistent performance with the smallest standard deviation.

While AAPM TG-178 recommends a ±3% tolerance for secondary dose calculations, our results suggest that tighter institutional tolerances may be achievable for routine LGK radiosurgery using any of the three software platforms evaluated. The perfect compliance of all secondary software platforms at the ±5% action level recommended by AAPM TG-178 confirms that this threshold provides an appropriate safety net for identifying major discrepancies requiring immediate intervention. Our data suggests that the ± 3% tolerance level serves as a more practical working threshold for routine quality assurance. For clinics setting up institution-specific levels, it should be noted that the tolerance levels should account for plan complexity, target location, and software-specific performance characteristics to avoid excessive false-positive investigations.

The clinical implementation of secondary dose calculation software extends beyond dosimetric accuracy to encompass workflow efficiency and user experience. All three software platforms can query data for calculations directly from the LGP database, eliminating manual data entry and reducing the risk of transcription errors. The user interface design was intuitive across all three software systems, with minimal training requirements for qualified medical physicists.

Skull geometry reconstruction methods, which can vary between simulated skull scaling and image-based approaches, can also represent a significant source of calculation variability, particularly for targets located near the skull surface.[28,29] For all the cases used in this study, the skull was rendered using the MRI dataset. The outliers observed on MUCheck are mostly associated with shallow targets near the skull surface, consistent with the previous study by Xu et al.[23] This suggests that for superficial targets, additional scrutiny or tighter action levels may be warranted when using MUCheck. In contrast, neither RadCalc nor ClearCalc showed systematic deviations for shallow targets potentially due to differences in skull reconstruction algorithms. This study employed a consistent MRI-based skull rendering approach across all patients; inter-user variability arising from manual skull delineation was not evaluated and represents a limitation warranting future investigation.

While all the frame cases in this study used the standard 90° gamma angle, cases replanned with the non-standard angles (70° or 110°), revealed platform-specific sensitivities. MUCheck showed increased calculation uncertainty for non-standard gamma angles compared to standard 90° gamma angle. In contrast, RadCalc and ClearCalc showed consistent performance across all gamma angles. Both RadCalc and ClearCalc maintained 100% compliance at ± 3% regardless of patient positioning, but MUCheck had cases exceeding the ± 5% action level recommended by AAPM TG-178. This finding should be considered when establishing institution-specific action levels, particularly for centers that frequently use non-standard gamma angles.

One of the limitations of this study is the exclusive focus on TMR10-based calculations in LGP TPS, which represents the current clinical practice, but excludes evaluation of the convolution-based algorithm. The TMR10 algorithm assumes water-equivalent tissue and does not account for tissue heterogeneities, which may be clinically significant for targets near bone interfaces or other density variations.[10,11,12,13] Future studies should investigate software performance with convolution-based primary calculation to assess the validity of secondary calculation approaches.

While our patient cohort was diverse in terms of target sizes and locations, the distribution of gamma angles for the frame cohort was skewed toward the standard 90°, potentially limiting generalizability to centers with different clinical practices. Institutions that routinely employ non-standard gamma angles may observe different performance characteristics. A study with a more balanced representation of gamma angles would provide more robust conclusions about platform performance across the full range of clinical positioning scenarios.

Finally, our study did not evaluate the impact of sector blocking patterns on calculation accuracy. Complex cases with extensive sector blocking to avoid critical structures or shape dose distributions may present additional challenges for secondary calculation systems. While our cohort of patients included plans that had some shots with sector blocking, future work should examine the performance for highly modulated treatments with non-standard sector configurations to ensure that the platforms maintain accuracy under these challenging conditions.

While the high compliance rates observed across all three software platforms reflect not only the accuracy of the secondary dose check but also the inherent precision of the LGK delivery system, independent secondary verification serves an important role beyond detecting large dosimetric discrepancies. It provides a systematic safeguard against software commissioning errors, data transfer failures, and institution-specific workflow deviations. AAPM TG-178 recommends a tolerance level of 3% and action level of 5% for an independent check of dose calculation for LGK radiosurgery with the expectation that exceeded action levels prompt review of input data in both the TPS and the independent software. The value of independent verification lies in detecting the rare but clinically consequential outlier.

CONCLUSION

Evaluation of the three commercial secondary dose calculation software platforms for LGK radiosurgery demonstrates that MUCheck, RadCalc, and ClearCalc all provide clinically acceptable dose verification capabilities. While statistically significant differences exist among the platforms when compared to the LGP TPS, all deviations remain within established clinical tolerance levels. Clinically, the use of independent secondary dose verification should be an integral part of quality assurance in LGK radiosurgery. Ensuring that planned and delivered doses are within acceptable variance is critical for patient safety and treatment efficacy.[14,15] All three software programs evaluated in this study provide valid secondary checks, with the choice of software dependent on institutional workflow requirements.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.