Couch Displacement Effects on Volumetric Modulated Arc Therapy Delivery and Verification of Simplified Couch Structure

*1)Department of Radiation Oncology, Juntendo University Faculty of Medicine, Tokyo, Japan, *2)Department of Radiology, Juntendo University Nerima Hospital, Tokyo, Japan, *3)Hiroshima High-Precision Radiation Cancer Center, Hiroshima, Japan, *4)Department of Radiation Oncology, Institute of Biomedical & Health Sciences, Hiroshima University, Hiroshima, Japan, *5)Department of Radiation Oncology, Tokyo Womenʼs Medical University, Tokyo, Japan


Introduction
Carbon fiber couches are in common use as patient support devices in radiotherapy since they are light and strong 1) .Also, carbon fiber should produce minimal absorption properties when beams pass through the couch before entering the patient, and it can permit treatment of the patient at multiple beam angles.Several authors have studied the dosimetric effect of carbon fiber couch attenuation 2)- 19) .The couch attenuation leads to a difference between the dose prescribed and the dose received by patients during posterior irradiation, and the magnitude of the attenuation is different for different energies and field sizes.In particular, for small fields, the attenuation effect can be enhanced compared to large field sizes because of the effect of scatter contribution.Njeh and colleagues showed that the effects of couch attenuation for 6 MV photons at 180 °were 4.9% and 3.4% for 5×5 and 10×10 cm 2 field sizes, respectively 3) .This characteristic is important for advanced techniques, such as intensity modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT), since a number of small segments can pass through the couch structure.For a RapidArc VMAT treatment where the couch structure is not included in the dose calculation, Popple and colleagues showed that up to 5.8% dose difference between the calculation and measurement was caused at the isocenter 4) .According to other studies, the measured doses were 3.2% and 4.2% lower than the calculated dose that did not include the structures of the couch and rails (rails-out) for RapidArc VMAT and multi-beam fixed gantry IMRT, respectively 5) , and up to 2.6% difference between the VMAT dose calculation and the ion chamber measurement was observed 6) .This underdosage can cause a reduction in planning target volume (PTV) coverage 5) 7) .Therefore, it is possible that a couch attenuation effect should be taken into account in treatment planning in order to estimate its effect on dose correctly 8) .For this purpose, a computed tomography (CT) couch top is replaced with the actual treatment couch top installed on the treatment machine, or a CT scan of the treatment couch is fused with the planning CT 9) .An alternative method is to construct a geometric model of the carbon fiber couch within the treatment planning system (TPS) 7) 10) .
The attenuation effect can have angular dependence with the gantry angle since the path length for the beams in the couch changes.This can result in variable dosimetric effects due to the couch structure if the position of the patient relative to the couch varies day to day.Ideally, the immobilization device is placed in the same position on the couch and patients are treated using the same geometry as the CT scan.However, such ideal conditions are not always met except for couches that permit firm fixation of the immobilization device to the couch top and thus the patient can lay on that exactly the same as for the CT scan.Therefore, it is necessary to evaluate the couch attenuation effects on the patient taking into account the differences in patient positions relative to the couch.Such couch displacement effects have been examined in the dose of a single field with a field size of 10×10 cm 2 10) .For the complex fields used in IMRT and VMAT deliveries, the effects on head and neck cases have been discussed 11) .
In this work we evaluated the couch-attenuation effects on a commercial Monte Carlo TPS (Monaco version 2.03, Elekta AB, Stockholm, Sweden) and discuss how the couch displacement affects dose distribution of prostate VMAT treatment.Also, we examined the availability of a simplified couch structure that keeps the dosimetric properties unchanged.

Couch modeling in TPS
The iBEAM evo carbon fiber couch (Elekta AB, Stockholm, Sweden) is constructed from a lowdensity foam interior and two layers of carbon fiber 12) .A cross section of the couch is shown in Figure -1.In order to assess the couch geometry, the couch set was cone-beam CT scanned and DICOM exported to the commercial IMRT TPS Monaco version 2.03 (Elekta AB, Stockholm, Sweden).The headrest part of the couch was scanned on a GE HiSpeed (NX/i S) CT scanner (GE Medical Systems, Milwaukee, WI) with an axial slice thickness of 0.2 cm to evaluate the actual CT numbers.Elekta quote electron densities of 0.0750±0.0005electron density of water (ρe.w) for the inner foam and 1.7±0.1 ρe.w for the carbon fiber layers 12) .Therefore, the couch outer layer was assigned an electron density of 1.7 ρe.w and the inner foam electron density a 0.075 ρe.w based on the manufacturersʼ specifications and the actual CT numbers of the couch headrest.In VMAT planning, the above couch model was loaded and moved to the proper position under the patient or in phantom CT images in order to take the actual couch attenuation effect into consideration for dose calculations.
The attenuation measurement with a simple square-field, 5×5 cm 2 , was used for the verification

Lateral couch displacement
The lateral position of a prostate patient may have a day-to-day variation of up to few cm, even if the patient is positioned with some immobilization devices.Table-1 shows the values of the couch lateral shifts for five prostate VMAT patients treated in our hospital where only the BlueBAG TM (Elekta AB, Stockholm, Sweden) was used as an immobilization device.The values of couch lateral shifts were obtained as the difference between the original couch position and the couch position after correction for patient position by cone-beam CT, the isocenter position.The maximum and mean values of lateral shifts for our prostate patients were observed around 5.0 cm and 1.5 cm, respectively.In this work, we simulated the dosimetric effect of variation in couch position relative to the isocenter for VMAT plans by moving the couch with respect to the CT slice center in a lateral direction as shown in Figure -2.We calculated the dose distributions for the couch position of 2 and 5 cm shifts relative to the isocenter and compared them to that of the original position.

Ion chamber and film measurements
Measurements were performed using a 6 MV photon beam of the Elekta Synergy linear accelerator (Elekta AB, Stockholm, Sweden) in order to evaluate the accuracy of the couch model in Monaco and the dosimetric effect of couch lateral shifts.Point dose measurements were performed using a 0.6 cc PTW Farmer chamber (PTW, Freiburg, Germany) and a 0.125 cc PTW semiflex ionization chamber (PTW, Freiburg, Germany).Isodose distributions for VMAT plans were measured using EBT2 Gafchromic film (International Specialty Products, Wayne, NJ, US) and compared to Monaco computational results by MapCHECK software (Sun Nuclear Corporation, Melbourne, FL, US).EBT2 films were sandwiched between 40×40×20 cm 3 tough water phantoms (Kyoto Kagaku Corporation, Japan) at a 10 cm depth with a source-tofilm distance (SFD) of 100 cm.A commercial flatbed scanner (EPSON GT-X970, EPSON, Nagano, Japan) was used to digitize the film at a resolution of 75 dpi using MapCHECK software.Comparisons between films and calculations were performed at 3% dose discrepancy and 3 mm distance-to-agreement (DTA) criteria with a threshold of 50%.

Prostate VMAT plan
Five prostate VMAT plans were generated for gantry angles from -180 °to 180 °in 10 °increments for a beam energy of 6 MV and we evaluated the effect of lateral couch displacement.Minimum segment area and segment MU were set to 2 cm 2 and 5 MU, respectively.A Monte Carlo dose calculation algorithm in Monaco was used for all of the VMAT plans with a grid size of 2 mm and variance of 2%.

Simplified couch modeling in TPS
It can be complicated to model the exact couch structure in TPS.We thus examined a rectangular couch shape keeping the radiological path length almost same as that of the original couch.The dimensions of the rectangular couch shape and relative electron density to water was 6.0×530 mm 2 with a uniform electron density of 1.0 ρ e.w .We applied the simplified couch modeling to a simple square-field, 5×5 cm 2 , and prostate VMAT plans, and evaluated its validity.

Verification of couch modeling in Monaco
Prior to the comparison of VMAT treatment plans, attenuation measurements with a 5×5 cm 2 field were performed for the verification of the couch model in Monaco TPS.calculation without couch correction had a deviation from 3.0% to 6.9% in dose (underdose) at the isocenter.The largest discrepancies were observed in the gantry angles of 120 °and 240 °, where the path length of the beam through the couch is longer and the magnitude of attenuation is larger compared to other gantry angles.For all gantry angles, these discrepancies are decreased to 1.3% after taking into account the couch transmission.This confirms that the combination of the couch dimensions and CT values in each couch layers adopted in this work are reasonable to correct the couch effects in Monaco TPS.We then used this couch modeling for VMAT treatment planning.The effects of couch correction for prostate VMAT plans were measured with the ionization chambers and EBT2 film.As shown in Table-3, the dose differences between the measurements and calculations are improved by around 2% by taking into account the couch attenuation.We performed a comparison of the measurements by films and Monaco calculated dose distributions in Map-CHECK software.For tolerances of 3% dose difference and a 3 mm distance to agreement (denoted 3%/3 mm DTA), we obtained more than a 95% passing rate of relative dose comparison.These results demonstrate a good agreement between film measurements and calculated results, taking into account the couch attenuation effects in the TPS.

Dosimetric effect of variation in couch position
Using the couch model, we evaluated the effects of couch shift on VMAT plans.Table-4 shows the couch shift effects on dose to isocenter for five prostate VMAT plans in the Monaco simulation.All of the deviations are less than 0.50% even though the couch is shifted by 5 cm laterally.The averaged deviations are 0.28% and 0.34% for the 2 cm and 5 cm shifted cases, respectively.The couch shift effects on dose distributions for coronal and sagittal planes are summarized in Table-5, where the comparison was performed under the conditions of 1%/1 mm, 2%/2 mm, and 3%/3 mm DTA and 10% threshold.The passing rate is more than 97% when the criteria are 1%/1 mm DTA.Figures -3A and 3B show one of the simulation results for 2 cm and 5 cm shifts relative to the original position, respectively.The comparison was performed under 1%/1 mm DTA and threshold of 10% criteria.As shown in the left bottom panels, both for Figure-3A and 3B, there are no deviations observed in high dose (target) regions.Most points where the deviations are larger than the criteria are found in the low dose region.The fail points are observed in regions outside the target, the same as the results for coronal planes.
We also evaluated the couch shift effects with EBT2 film measurements.Figure -4 shows the comparison of the measured planar dose in the original phantom location and that of the 5 cm shifted plan.The comparison was performed with the criteria of 3%/3 mm DTA and threshold of 50% due to the dosimetric limitation of film measurements.Some points deviating from the criteria are

Evaluation of simplified couch modeling
The phantom position relative to the couch did not cause any significant discrepancy in the dose distribution.This means that the shape of the iBEAM evo carbon fiber couch does not have any importance when discussing the couch attenuation effect on patient dose.This result indicates that the couch shape can be simplified only to take the attenuation effect into consideration appropriately.

cm shift cm shift Original
Difference Difference

Figure-4
Film measurements comparison between the original phantom position and 5 cm shift.The comparison is done with 3% /3 mm criteria and threshold of 10%.

Rectangular shape Original couch
Difference

Figure-5
Comparison of the planar dose distributions between the original couch and rectangular shapes.
VMAT plan Couch shift (cm) 0 cm 2 cm 5 cm We compared the doses at the isocenter with a field size of 5×5 cm 2 with the original couch and the simplified shape, 6.0×530 mm 2 , from 120 °to 240 °.For all of the gantry angles, the difference was less than 1.0%.For prostate VMAT plans, we found that the isocenter dose differences between the original and shifted plans were less than 1.0% as shown in Table-6. Figure -5 shows the planar dose comparison between the rectangular shape and the original shape with criteria of 1%/1 mm DTA and threshold of 10%.The passing rate obtained was 99.2% and most of the deviations were outside the high dose region.Therefore, the dosimetric properties of the rectangular and the original couches are almost identical.

Discussion
In this study, we evaluated couch modeling in the Monaco TPS and demonstrated couch attenuation effects on VMAT planning doses for prostate patients.Gerig and colleagues 13) reported that the effect of the iBEAM evo carbon fiber couch on beam attenuation for a 10×10 cm 2 field size was about 2.5% in its central region for 6 MV.Work published by Smith and colleagues also examined dosimetric properties of the iBEAM evo couch 12) .They reported that ion chamber measurements showed beam attenuation to increase from 2.7% to a maximum of 4.6% for a 6 MV beam.Considering the field size dependence, our results are consistent with these previous studies.Smith and colleagues also investigated the surface dose enhancement characteristics of the carbon fiber couch.A limitation of this study is that we do not discuss the dose increase at and near surface of our phantom, which is needed as a future work.
We have also presented cases in which phantom positions relative to the couch changed from the original position assuming that the patient was shifted relative to the planned couch position.The effect of couch shifts of up to 5 cm relative to the original position are small and observed mainly outside of the target.There are no deviations found in the high dose region.Therefore, we believe that the effects of couch shifts do not degrade the PTV coverage.Since the discrepancies of dose distributions from the original and shifted plans observed in low dose regions are not systematically high or low (seem to be random errors), they may partly be attributed to the calculation uncertainty inherent in Monaco TPS.
Several studies showed that the patient setup error with some immobilization devices, which fix the patient position with respect to the couch, can be reduced to less than 1 cm (0.51 cm and 0.28 cm as total vector error for a device surrounding abdominal and pelvic region and a whole-body immobilization device, respectively 20) ).Our findings indicate that the couch displacement has only a small influence on the prostate VMAT plan even though the positioning systems indexed to the treatment couch were not used.Work recently published by Gardner and colleagues 14) showed the sensitivity to lateral shifts of other couch type, QFix kVue Calypso-compatible couch top with rails which was composed of a periodic lattice structure (250 HU, 1.099 ρ e.w ) in order to minimize imaging artifacts.They concluded that the largest sensitivity to lateral shifts occurred at the lateral edge of the rail structures and the mean magnitude of the dose deviation was 0.61% for 2 mm shifts, 0.46% for 5 mm shifts, and 0.86% for 10 mm shifts 14) .Therefore, the magnitude of couch lateral shifts depends on the couch structures.The iBEAM evo carbon fiber couch consists of low-density inner foam and a carbon fiber exterior and it plays an important role not only in minimizing imaging artifacts for cone-beam CT and orthogonal x-rays but also in minimizing effects on patient dose for couch displacement.
Insensitivity to the couch structure implies that the couch structure can be simplified for couch modeling in TPS.We thus simulated a simple couch structure, a rectangular shape whose dimension was 6.0×30 mm 2 with 1.0 ρ e.w in Monaco and compared the isodose distributions with that of the original shape.The passing rate with 1%/1 mm DTA and threshold of 10% criteria was 99.2% and it can be concluded that the simple couch shape is validated in Monaco TPS.The simple couch structure can ease the couch modeling in Monaco TPS and be helpful for procedures of actual patient planning.In this work we only discussed the prostate VMAT plans.For systematic future studies, it is necessary to evaluate other treatment sites, such as the head and neck and lung by using the same procedure.

Conclusion
When the couch dimensions were inserted into the TPS and the appropriate densities were assigned to those contours, the couch attenuation properties were modeled well.The dosimetric effects of variation in couch position were small and only seen in the region far from the target.Our findings for the phantom lateral displacement suggest that the positioning accuracy will have no clinical significance.We also found that the couch dimension can be simplified in TPS for the dose calculation.

Figure- 1
Figure-1 Cross section of the iBEAM evo carbon fiber couch in Monaco treatment planning system

Figure- 2
Figure-2Patient position relative to couch at original (A) and shifted positions (B), and its simulation in the Monaco treatment planning system (C) and (D), respectively.(C) Phantom is at the original position, and (D) shifted by 5 cm from the original position.
Figure-3Simulation of couch shift in the Monaco treatment planning system.(A) A 2 cm shift relative to the original position comparison and (B) a 5 cm shift comparison.The comparison is done with 1%/1 mm and a threshold of 10%.The differences between the shifted and original dose distributions are shown in the left bottom frames, where blue and red points represent the deviated points from the criteria.

Table - 1
Variations of couch lateral shifts for prostate VMAT patients during 38 fractions

Table - 2
Table-2 shows the couch correction effects for 6 MV photons with the gantry angles for the simple square-field.The Couch attenuation measurements compared to the Monaco calculation for 6 MV, 5×5 cm 2 fields.Calculated doses with/without couch correction in Monaco are shown.

Table - 3
Couch correction effects on doses at the isocenter in prostate VMAT treatment plans

Table - 4
Couch shift effects on the dose to the isocenter for five prostate VMAT plans.The discrepancy is expressed as the dose difference between the original and shifted plans.

Table - 5
Passing rates for the comparison between the original and couch shifted VMAT plans.The comparison was performed by MapCHECK software with criteria of 1%/1 mm, 2%/2 mm and 3%/3 mm DTA.

Table - 6
Evaluation of couch shifted effects on the dose to the isocenter for a simplified couch model