Evaluation and verification of the QFix EncompassTM couch insert for intracranial stereotactic radiosurgery

Abstract The QFix EncompassTM stereotactic radiosurgery (SRS) immobilization system consists of a thermoplastic mask that attaches to the couch insert to immobilize patients treated with intracranial SRS. This study evaluates the dosimetric impact and verifies a vendor provided treatment planning system (TPS) model in the Eclipse TPS. A thermoplastic mask was constructed for a Lucy 3D phantom, and was scanned with and without the EncompassTM system. Attenuation measurements were performed in the Lucy phantom with and without the insert using a pinpoint ion chamber for energies of 6xFFF, 10xFFF and 6X, with three field sizes (2 × 2, 4 × 4, and 6 × 6 cm2). The measurements were compared to two sets of calculations. The first set utilized the vendor provided Encompass TPS model (EncompassTPS), which consists of two structures: the Encompass and Encompass base structure. Three HU values for the Encompass (200, 300, 400) and Encompass Base (−600, −500, −400) structures were evaluated. The second set of calculations consists of the Encompass insert included in the external body contour (EncompassEXT) for dose calculation. The average measured percent attenuation in the posterior region of the insert ranged from 3.4%–3.8% for the 6xFFF beam, 2.9%–3.4% for the 10xFFF, and 3.3%–3.6% for the 6X beam. The maximum attenuation occurred at the region where the mask attaches to the insert, where attenuation up to 17% was measured for a 6xFFF beam. The difference between measured and calculated attenuation with either the EncompassEXT or EncompassTPS approach was within 0.5%. HU values in the EncompassTPS model that provided the best agreement with measurement was 400 for the Encompass structure and −400 for the Encompass base structure. Significant attenuation was observed at the area where the mask attaches to the insert. Larger differences can be observed when using few static beams compared to rotational treatment techniques.


| INTRODUCTION
Intracranial stereotactic radiosurgery (SRS) is a treatment technique used to deliver large doses of radiation to small targets in the cranium in order to manage primary brain tumors, metastasis, or functional diseases. Frameless mask-based systems have become popular over the past decade since they are noninvasive; allowing for greater patient comfort as well as the ability to fractionate treatments while still retaining the immobilization accuracy of frame-based treatments. [1][2][3] Current frameless-based systems typically use a clam shell style mask to immobilize the patient in order to provide submillimeter accuracy treatments to small intracranial lesions. 4 Frameless systems use either extensions in which the mask system extends off the patient support structure, or overlays in which the mask system is attached and indexed to the carbon fiber patient support structure. The QFix Encompass TM SRS immobilization system, created by QFix (Avondale, PA, USA) consists of a couch insert, and a thermoplastic mask attached to the raised component of the insert. The geometry and design of the insert is unique in that high density carbon fiber material surrounds the cranium, which may interfere with the target area to be treated.
Several groups have demonstrated the importance of modeling immobilization devices in the treatment planning system (TPS) to limit their dosimetric impact, particularly on skin dose, dose distribution, and attenuation. [5][6][7][8][9][10][11] A TPS model of the QFix Encompass insert has been created and is available in the Eclipse TPS software, v15.5 (Varian Medical System, Palo Alto, CA, USA). The QFix Encompass immobilization device is an integral part of the Varian HyperArc TM High-definition radiotherapy automated SRS delivery workflow. The immobilization device allows the patient to be located in space relative to the machine isocenter to ensure machine clearance and efficiency during automated delivery. In this study, we evaluate the dosimetric properties of the QFix Encompass TM system and quantify the amount of attenuation through the system. The Hounsfield Unit (HU) values of the couch model in the TPS were verified. Finally, we evaluated the dosimetric consequences and robustness of the system.

2.A | The QFix Encompass TM model
The QFix Encompass TM SRS immobilization system consists of two parts: the Encompass insert and the clam shell style Fiberplast TM mask [ Fig. 1(a)]. The Encompass insert is an immobilization device that can be attached to or overlaid on the treatment couch. The clam shell style mask consists of an anterior and posterior portion which is customized for each patient during simulation. The Fiberplast TM mask is a low temperature thermoplastic that hardens quickly, typically within 10 min. The mask is aligned to the insert with acrylic pins and locked into place with adjustable shims.
To evaluate the CT numbers of the Encompass system, a mask was made on the Lucy 3D QA phantom (Standard Imaging, Middleton WI, USA). After the mask hardened, a CT scan was acquired of the mask and Encompass insert using a Philips Brilliance Big bore scanner (Philips, Netherlands), using our institution's intracranial SRS protocol. (120 kVp, 400 mAs, 1 mm slice thickness, FOV = 350 mm, 512 9 512 Matrix). The HU values were evaluated for all portions of the Encompass system.
The QFix Encompass TM insert is modeled in Eclipse TPS v15.5 as a support structure. The Encompass TPS model is a simplified model of the full Encompass system that does not include some portions of the couch. The Encompass TPS model consists of two separate structures: Encompass and Encompass Base. The "Encompass" structure includes the bulk portion of the carbon fiber U-shaped insert.
The "Encompass Base" structure includes the posterior region of the insert system and is made up of a double-layered section with a hollow interior [ Fig. 1(b)].

2.B | Phantom setup
Measurements were performed in a spherical, Lucy 3D QA phantom (Standard Imaging, Middleton WI, USA) with a PTW Pinpoint ion chamber (Freiburg, Germany), 0.015 cc active volume. The Lucy phantom was immobilized by creating a custom mask in the Encompass insert system [ Fig. 1(a)]. A CT scan of the phantom and Encompass insert was acquired using the SRS protocol. The scan was imported into the TPS, where two image sets were generated based on how the Encompass system was to be included for dose calculation.
The first image set was the Encompass TPS image set (Encompass TPS ) which was contoured according to the vendor recommendations for incorporating the Encompass couch structure onto a patient image set. This consists of contouring the entire Encompass system in the external body contour, including the patient and mask. The Encompass TPS model is inserted as a support structure, and then removed from the external body contour using the Boolean tool. The contouring procedure results in the external body contour encompassing the patient and mask, and the Encompass insert as a separate support structure [ Fig. 1(d)]. This allows the dose calculation to take into account portions of the mask system that are custom to each patient and are not included in the couch structure.

| 223
Encompass insert to quantify the amount of attenuation, which was then compared with calculated values from the three image sets above.

2.C | Attenuation measurements
Measurements were performed on a Varian EDGE linear accelerator (Varian, Palo Alto, CA, USA) using three energies: a flattened 6 MV beam (6X), and flattening filter free 6 MV (6xFFF) and 10 MV (10xFFF) beams. The measurements were performed for each energy, with three field sizes: 2 9 2, 4 9 4, and 6 9 6 cm 2 . A total of 18 measurement setups were performed with and without the Encompass insert. The Encompass and Lucy setups were aligned using CBCT, matched to the TPS CT with 6 degrees of freedom.
A total of 41 measurements per energy and field size were performed. The measurements were broken down into four zones (Fig. 2). Zone 1 (blue) is a 70°area that represents the area of the "Encompass

TPS Model
The HU values for the Encompass TPS model were determined for each of the structures in the Encompass model by choosing the HU that minimized the difference between the measured and calculated percent attenuation in a specific zone. The HU value for the "Encompass Base" structure was determined from measurements in Zone 1, the region of double-layered carbon fiber. The measurements were averaged and compared to the structure set HU values of À600, À500, and À400. Similarly, the HU value for the "Encompass" structure was determined by choosing the HU that minimized the differ- To verify that all portions of the Encompass system were correctly modeled in the Encompass TPS model, the percent attenuation was also calculated with the Encompass TPS and compared to the percent attenuation calculated with the Encompass EXT image set.

2.E | Clinical case recalculation and measurement
Ten clinical cases were recalculated with the Encompass TPS model (Encompass TPS ) and compared to the clinical plan, where the Encompass system was taken into account in the external contour (Encompass EXT ). The targets in the ten clinical cases ranged in location as well as in size. The treatment techniques included volumetric modulated arc therapy (VMAT) and dynamic conformal arc (DCA); both techniques are implemented for SRS treatments at our institution. Table 1 summarizes the lesion location, size, and treatment technique for the 10 clinical cases recalculated.
The dose to 99% (PTV D99% ), 95% (PTV D95% ), and 0.035 cc (PTV D0.035cc ) of the PTV was evaluated between the clinical plan and the same plan recalculated with the Encompass TPS model. For the highest priority organ-at-risk, the dose to 0.035 cc (OAR D0.035cc ) was evaluated. For the acoustic neuroma cases, the OAR with the highest priority is the cochlea. However, for a solitary metastasis with no physiological OARs within a 2 cm radius, a 0.5 cm ring around the PTV was created to simulate an OAR around the lesion. A paired student t-test was used to evaluate the differences in PTV and OAR dose between the two plans, where P < 0.05 was the threshold for statistical significance.

3.B | HU validation
The HU values of the components of the Encompass systems are summarized in Table 3. The final HU value chosen for the Encompass TPS was 400HU for the Encompass structure and À400 for the Encompass base structure. This minimized the difference between measured and calculated attenuation, (Atten Calc -Atten Measured ). The percent difference between the measured and calculated attenuation for the Encompass TPS for HU evaluation is summarized in Table 4.
In Zones 1-3, the average difference between measured, Atten Measured, and calculated attenuation, Atten Calc, for the Encompass TPS for the 6xFFF beam was À0.3%, À0.2%, and 0.0%, for the 10xFFF beam 0.2%, 0.2%, and 0.3% and for 6X beam 0.1%, 0.2%, and 0.4%, for field sizes of 2 9 2 cm 2 , 4 9 4 cm 2 , and 6 9 6 cm 2 , respectively. The maximum difference between measured and calculated attenuation with the Encompass TPS was 3.0% which occurred in Zone4 for the 6X beam for a 6 9 6 cm 2 field size. The percent differences are relatively similar in range to those obtained when dose was calculated from HU values obtained directly from the CT in the Encompass EXT calculations. Figure 2(b) shows the measured and calculated percent attenuation from the Encompass EXT and Encompass TPS .

3.C | Clinical case recalculation and measurement
The average difference in PTV coverage between the patient data sets including the Encompass insert in the external with the TPS T A B L E 1 Summary of location, number of fractions, total dose, treatment technique and target volume for the 10 clinical cases in the study that were recalculated. Two plans treated two targets simultaneously, and the target volume for each target is shown. Percent difference in isocenter dose between measured and calculated with only Lucy (Lucy only ) and with Lucy in the Encompass system (Lucy ENC are summarized). T A B L E 2 Summary of the percent attenuation measured using a pinpoint ion chamber for 6X, 6xFFF, and 10xFFF photon energies for field sizes of 2 9 2 cm 2 , 4 9 4 cm 2 , and 6 9 6 cm 2 . The average percent attenuation (minimum, maximum) values are shown for Zones 1-3 and separately for Zone 4 where more attenuation is observed.

3.D | Couch placement sensitivity
In With a 3 mm translation displacement of the Encompass insert TPS model, the average difference for the clinical case recalculation for PTV D99% , PTV D95% , PTV D0.035 cc , and OAR D0.035 cc was 0.02%, 0.02%, 0.003%, and À0.1%, respectively. The average percent difference at isocenter for the three treatment beams was 0.04%, 0.00%, and 0.07% for beam 1 at couch 0°, beam 2 at couch 280°and beam 3 at couch 40°, respectively. The maximum difference occurred at beam3, couch 40°when the couch was shifted 3 mm laterally. The magnitude of calculated attenuation between the Encompass EXT and Encompass TPS was similar. Differences in Zones 1 can be attributed to the way the Encompass base is modeled in the TPS model (Fig. 2). The base layer of the Encompass insert consists of a double layer of high-density carbon fiber with a hollow center. The base layer was modeled as a solid piece due to the difficulty of modeling a thin layer of material in the TPS and transferring the T A B L E 3 Summary of components of the QFix Encompass TM SRS immobilization system, corresponding HU value ranges, and whether the component is included in the Encompass TPS model.  Depending on the couch angle, the change in isocenter dose varies between beams. During the SRS planning process, arc geometry is typically chosen to achieve conformal dose distributions rather than to avoid portions of the mask that are more attenuating. However, for static fields, such as IMRT or 3D conformal techniques, care should be taken to avoid areas of high attenuation since the dosimetric consequences can be accentuated. Figure 3(b) demonstrates an example of whole brain opposed lateral radiotherapy treatment for a patient initially simulated for an SRS treatment. Areas of high attenuation occur at the clip area, resulting in decreased coverage to the brain. The patient was ultimately treated with whole brain using a hippocampal sparing VMAT technique.

HU Range Included in TPS Model
In this study, the anisotropic analytical algorithm (AAA) was used to verify the HU values of the TPS model. For AAA, inaccuracies in dose calculation have been demonstrated at the interfaces of materials. 13,14 In the Encompass mask system, dose inaccuracies can occur between the mask, insert, and air. However beyond the interface region, dose from AAA is often comparable to Monte Carlo based algorithms, within 2%-4%. 15 Other studies have also found that couch models included in the external body contour for dose calculation often agree with measurements within 2%. 5,9,11 A limitation of this study is that skin dose was not evaluated.
Because the mask is an integral part of the Encompass system and is customized for each patient, the mask acts as additional build up and the amount may vary from patient to patient. Also, for rotational type techniques such as DCA or VMAT, dose to the skin is often spread out across the arc path length. Furthermore, the largest dosimetric differences occurred at gantry 270 through the shimming system of the mask. The change in attenuation when the shimming level is adjusted was not evaluated, and may change as air gaps are introduced in the system. Future studies could be performed evaluating the dose to skin, as well as the impact of shimming level, static treatment fields as well as using more accurate algorithms such as Monte Carlo.

| CONCLUSION
Significant attenuation occurs when using the QFix Encompass TM SRS immobilization system, and occurs at the area where the mask attaches to the insert. HU values for the Encompass TPS model were found to be 400 for the Encompass structure and À400 for the Encompass base structure, which resulted in an average percent difference between measured and calculated attenuation of less than 0.5%. Small uncertainties in couch placement do not significantly perturb the dose calculation. However, larger differences can be seen when using few static beams compared to rotational treatment techniques.
F I G . 3. (a) Coronal cross-section of clinical patient recalculated with translational shifts demonstrating sensitivity of positioning of beams relative to high density portions of the Encompass insert (magenta). (b) Axial cross-section of two field, opposed lateral beams for whole brain radiotherapy treatment demonstrating areas of high attenuation through the clips resulting in decreased coverage to the brain.