"Tuesday 11 June"

Purpose: To develop a phase space source model enabling Monte Carlo (MC) dosimetry calculations and verification of Gamma Knife treatments in inhomogeneous geometries.

Materials and methods: A previously validated Gamma Knife Perfexion (GKPFX) MC-based detailed source model was used to create single sector phase-space (PHSP) source models for the three available collimators.  These were validated in terms of single sector and single shot X-Y-Z dose profiles in a spherical water phantom with corresponding data obtained using the detailed model and experimental EBT-3 film measurements.

The PHSP-source model was subsequently used to validate GammaPlan (LGP) dose predictions using the convolution algorithm for a plan using a composite shot consisting of all collimator sizes delivered in a virtual phantom of 8 cm radius containing a 1.5 cm thick hemispherical bone inhomogeneity in the vicinity of the 50% isodose line.

Results: Single sector and single shot dosimetry results using PHSP simulations were found in excellent agreement with corresponding detailed MC model calculations and film measurements. Indicatively, gamma passing rates above 99.5% were achieved for local 1%/1mm criteria against detailed model simulation and 2%/1mm criteria against film measurements. Efficiency gain by a factor of up to 2500 for the smaller field size was attained compared with detailed model simulations. Convolution absolute dose distribution evaluation using the PHSP-source model simulations in the inhomogeneous phantom resulted in a gamma passing rate of 99.15%, applying 1%/1mm local gamma index criteria and 1% dose threshold.

Conclusion: An accurate and efficient GKPFX single sector PHSP source model was developed and validated. LGP calculations using the convolution algorithm were evaluated in an inhomogeneous geometry using this model and found to be in excellent agreement, indicating the accuracy of the convolution algorithm in water-bone inhomogeneities. Further work on convolution algorithm verification on more complex and clinical cases is in progress.

The IAEA TRS-483 code of practice requires that solid state dosimeters used for quality assurance in small field radiotherapy be utilized in a “face-on” orientation [1]. However, this practice means that the high spatial resolution of the PTW microdiamond or uD in “edge-on” orientation is unrealized [2]. The aim of this study was to characterize the uD for small field applications in an edge-on orientation. To that end, the detector went through a rigorous characterization of its performance in both edge-on and face-on orientations for different field sizes and angular incidences

Output factor (OF), Percentage Depth Dose (PDD) curves and field profile measurements were performed with the uD in edge-on and face-on orientations and compared against the IBA RAZOR for 6MV photon field for both FF/FFF modalities in a IBA blue water phantom on a Varian True Beam linac for square field sizes between 0.5-10 cm. Angular dependence as a function of field size measurements (0.5x0.5-3x3cm²) were also performed in two different cylindrical PMMA phantoms to investigate the effect of orientation upon angular dependence.

The high spatial resolution of the uD in edge-on, allowed for precise profilometry of small FF/FFF square fields to be performed. The uD was shown to over-response in edge-on in comparison with face-on for fields ≤2x2cm². Angular dependence measurements in the cylindrical edge2face phantom showed a 6-12% difference in response of the uD in the edge-on and face-on orientations for 0.5-3cm square fields, although larger variations (~31%) were observed. Additional angular dependence measurements in the cylindrical edge2edge phantom shows that the uD is almost angular independent over a range of 180° with differences of ±1%.

In edge-on orientation, the uD was shown to be suitable for profile reconstruction as well as exhibiting negligible angular dependence (±1%) making it an option for specific clinical applications. However, the orientation is deemed to be unsuitable for PDD and OF measurements, due to a less than ideal build-up behaviour and over-response. Full results including that of a dedicated Monte Carlo simulation study to optimise the detector packaging will be presented at the ISRS congress.

[1] H. Palmans, et al, Technical Report Series No. 483. International Atomic Energy Agency, Vienna; 2017

[2] V. De Coste, et al, Phys. Med. Biol. 62 (2017) 7036-7055

Purpose/Objective:Stereotactic radiosurgery (SRS) is a promising treatment option for patients with 4 to 10 brain metastases (BM). We studied whether automated planning can improve LINAC-based stereotactic radiosurgery plan quality for multiple BM. 

Materials/Methods:For 12 patients with 4 to 10 BM, five non-coplanar LINAC-based SRS plans were created for 6MV photons: a manually planned dynamic conformal arc (DCA) plan with a separate isocenter for each metastasis, a dynamic IMRT plan with one isocenter, a VMAT plan with one isocenter, two DCA plans with one isocenter for three and five couch rotations. The last three plans were automatically generated. The prescription dose was 21Gy or 18Gy single fraction or 25.5Gy in 3 fractions depending on the volume of the largest metastasis and prescribed to the 80% isodose line.The PTV coverage should be at least 98%.To assess SRS plan quality, the Paddick conformity index (CI), the Paddick gradient index (GI), the total V12Gy and V5Gy and the number of monitor units (MU) were studied. 

Results: The mean CI was the highest for dynamic IMRT and manual DCA plans. The lowest GI was for manual DCA plans with a separate isocenter for each metastasis and for automatically generated DCA plans with one isocenter, the highest GI was for VMAT plans. The V12Gy of automatically generated DCA plans with one isocenter and dynamic IMRT plans were comparable with the manual DCA plans. The number of MU was the smallest for VMAT plans, followed by IMRT and automatically generated DCA plans.

Conclusions: Automatically generated LINAC-based, single isocenter SRS plans for multiple BM result in fewer MUs, with a plan quality comparable to manual multiple-isocenter DCA plans. Based on all compared parameters, dynamic IMRT and DCA plans with one isocenter were the best and comparable with multiple-isocenter DCA plans. 

Objectives: To evaluate inter- and intra-fraction motion detected using frameless immobilization for Gamma Knife (GK) stereotactic radiosurgery (SRS).

Materials and Methods: Following consent to frameless GK-SRS, patients were immobilized with a thermoplastic mask followed by acquisition of a reference CBCT scan. Daily setup verification and intra-fraction motions were monitored using CBCT and an intra-fractional motion management (IFMM) system.  Patient setup and CBCT was repeated when IFMM thresholds were exceeded or when the patient needed a voluntary break. In-house Matlab scripts were developed to parse log files to determine patient inter- and intra-fraction setup variability.

Results: Thirty-eight plans were reviewed from 36 patients (2 patients treated twice). The average number of targets per plan was 1.3 [range: 1-4] and treatment time was 42 min [range: 8.3 - 145.9min]. The number of CBCT per fraction is 1.8 [range: 1-7]. Systematic setup error was found by the difference between reference and daily CBCTs as 0.93, 1.17, 1.17 mm and 0.8, 0.6, 2.2 degrees in x, y, and z direction respectively. Random error (intra fraction) was found 0.40, 0.33, 0.35mm and 0.3, 0.3, 0.6 degrees from successive CBCTs. IFMM measurement with marker motion larger than 0.2mm are  77 times/min during beam delivery and the average directional motion during beam on was 0.0, -0.1, 0.3mm (standard deviation of 0.4, 0.4, and 0.6mm) in x, y, and z direction. Systematic (random) motion of IFMM was 0.7mm, 0.5mm, 0.9mm (0.3, 0.2, and 0.5mm).

Conclusions: Preliminary analysis suggests good setup reproducibility with the largest discrepancy in the z-direction. After setup correction, random intra-fraction motion was found to be within 0.5mm with larger systematic motions triggered for pause or correction by the IFMM.

Purpose: Adapting manufacturer’s end-to-end test to the Gamma Knife Icon mask system, we were able to verify the accuracy of position correction in Gammaplan even for large angular and translational shifts. However, the test does not verify if isodose volume is preserved.

Methods and Materials: An anthropomorphic head phantom with a film insert in the mid-coronal plane is used. Lesion-E has an elliptical shape covered by one single composite shot. Lesion-S has a sausage shape covered by 4 composite shots. Close to either lesion are organs at risk (OAR1 and OAR2). For each lesion, a non-composite plan was also created to produce similar prescription isodose volume with comparable dose to OARs. The phantom was treated in the planning position (A), and in a position shifted 4 cm superiorly and rotated 95 degrees to right (D). For lesion-S, the phantom was irradiated in two additional positions: 14-degree chin-up (B), and 14-degree rotation to right with 7-degree chin-up (C). (Min, max, mean) dose reported under dose evaluation during treatment were analyzed. Gamma Index comparison of film dose at positions B, C, or D versus A was used. Prescription dose was 3 Gy per fraction.

Results: Non-composite-shot plans: All Gamma Index passing rates are > 97%, and all differences in (min, max mean) dose are <= 0.1 Gy. Composite-shot plans: Passing rate is 57% for position D for lesion-E, and 92%, 78%, and 44% for position B, C, and D, respectively for lesion-S. The difference in (min, max, mean) doses becomes larger as the phantom shifted from position B through D: from a maximum 0.4-Gy difference in position-B to a maximum difference of 0.8-Gy in position-D for lesion-S and as large as 1.4 Gy for lesion-E.

Conclusions: For robust planning, it is recommended to use only non-composite shots for mask-based treatments with Icon.

Objective:

Dosimetric and geometric accuracy are paramount in Stereotactic Radiosurgery (SRS) to achieve effective and safe implementation of the treatment. In this work, End-to-End accuracy was evaluated for single-isocenter multi-focal SRS treatments in six centers.   

Methods:

Eight identical 3D-printed head phantoms were constructed using the planning-CT dataset of a patient, simulating bone structures by a bone equivalent material. Six phantoms (one per clinic) were filled with 3D polymer gel, which simulates brain tissue and acts as a dosimeter in combination with an MR scanner, while the other two phantoms were filled with water and equipped with an ion chamber and a film insert, respectively. A single-isocenter plan using a 5-arc VMAT beam arrangement was created in Monaco Treatment Planning System (TPS). Six targets were adjusted to achieve a range of target sizes 6-25mm in diameter at various distances from the isocenter. Prescription dose was set to 8Gy and dose delivery was performed by the Elekta Versa HD-HDRS linear accelerator with HexaPOD system, following departments’ clinical SRS workflow. Point, 2D, and 3D dose values were obtained by ion chamber, film, and gel measurements, respectively. Geometric accuracy of all targets was evaluated by the comparison of 2D/3D relative dose distributions between measurements and TPS calculations. Dosimetric accuracy was verified by ion chamber measurements in one target.     

Results:

Excellent geometric agreement (<1mm) between TPS calculations and measurements was observed for the targets lying less than 4cm from the isocenter. For the targets with a distance from the isocenter greater than 4cm, the average difference from all sites was 0.8mm with a maximum discrepancy of 1.9 mm. Ion chamber measurements yielded an average difference of 1.2% ± 0.5% leading to a superb agreement within uncertainties.   

Conclusion:

The overall accuracy of single-isocenter multi-focal SRS treatments was found within acceptable limits for all clinics using a patient-specific End-to-End methodology.