|Year : 2016 | Volume
| Issue : 1 | Page : 158-161
Influence of monte carlo variance with fluence smoothing in VMAT treatment planning with Monaco TPS
B Sarkar1, A Manikandan2, M Nandy3, A Munshi4, P Sayan4, N Sujatha5
1 Department of Radiation Oncology, AMRI Hospitals, Kolkata, India
2 Department of Radiation Oncology, Narayana Hrudayala, Bangalore, India
3 Chemical Sciences Division, Saha Institute of Nuclear physics, Kolkata, India
4 Department of Radiation Oncology, Fortis Memorial Research Institute, Gurgaon, Haryana, India
5 Department of Radiation Oncology, Guntur Medical College, Andhra Pradesh, India
|Date of Web Publication||28-Apr-2016|
Department of Radiation Oncology, AMRI Hospitals, Kolkata
Source of Support: None, Conflict of Interest: None
Introduction: The study aimed to investigate the interplay between Monte Carlo Variance (MCV) and fluence smoothing factor (FSF) in volumetric modulated arc therapy treatment planning by using a sample set of complex treatment planning cases and a X-ray Voxel Monte Carlo–based treatment planning system equipped with tools to tune fluence smoothness as well as MCV. Materials and Methods: The dosimetric (dose to tumor volume, and organ at risk) and physical characteristic (treatment time, number of segments, and so on) of a set 45 treatment plans for all combinations of 1%, 3%, 5% MCV and 1, 3, 5 FSF were evaluated for five carcinoma esophagus cases under the study. Result: Increase in FSF reduce the treatment time. Variation of MCV and FSF gives a highest planning target volume (PTV), heart and lung dose variation of 3.6%, 12.8% and 4.3%, respectively. The heart dose variation was highest among all organs at risk. Highest variation of spinal cord dose was 0.6 Gy. Conclusion: Variation of MCV and FSF influences the organ at risk (OAR) doses significantly but not PTV coverage and dose homogeneity. Variation in FSF causes difference in dosimetric and physical parameters for the treatment plans but variation of MCV does not. MCV 3% or less do not improve the plan quality significantly (physical and clinical) compared with MCV greater than 3%. The use of MCV between 3% and 5% gives similar results as 1% with lesser calculation time. Minimally detected differences in plan quality suggest that the optimum FSF can be set between 3 and 5.
Keywords: Fluence smoothening factor, monte carlo, volumetric modulated arc therapy
|How to cite this article:|
Sarkar B, Manikandan A, Nandy M, Munshi A, Sayan P, Sujatha N. Influence of monte carlo variance with fluence smoothing in VMAT treatment planning with Monaco TPS. Indian J Cancer 2016;53:158-61
|How to cite this URL:|
Sarkar B, Manikandan A, Nandy M, Munshi A, Sayan P, Sujatha N. Influence of monte carlo variance with fluence smoothing in VMAT treatment planning with Monaco TPS. Indian J Cancer [serial online] 2016 [cited 2021 Jun 13];53:158-61. Available from: https://www.indianjcancer.com/text.asp?2016/53/1/158/180820
| » Introduction|| |
The concept of volumetric-modulated arc therapy (VMAT) has been described by many authors.,,, VMAT is a system for intensity-modulated radiation therapy (IMRT) treatment delivery that achieves high dose conformity by optimizing the dose rate, gantry speed, and dynamic multileaf collimator movement.,,,,, The potential of VMAT delivery is to reduce treatment time, significantly, compared with traditional IMRT. Different treatment planning studies ,, have been published, comparing VMAT and dynamic IMRT, conformal radiotherapy, or stereotactic radiotherapy treatments with regard to dosimetric plan quality, delivery time, and monitor units (MU), using either commercially available treatment planning systems (TPS) such as Eclipse (Varian Medical Systems, Palo Alto, CA), Smart Arc in Pinnacle 3 TPS (Philips Healthcare, Andover, MA), or in house TPS's as reported by Dobler et al. The desirable characteristics of dose calculation engine in radiation therapy are that the calculation be fast enough so that the treatment planning process can be completed in a clinically acceptable time frame; and second, that the result of the dose calculation be sufficiently accurate. The uncertainty in the dose calculated by a conventional dose calculation algorithm has been reported to be between 5% to 10% in the presence of heterogeneities. Similar error values were also reported for dose calculated using Monte Carlo methods., CMS Monaco ® (Norcross, GA) V2.3 is radiotherapy TPS, which uses XVMC (X-ray Voxel Monte Carlo) code for dose calculation. This is a faster method with appreciable accuracy of 2% when compared with the experimental data in water phantom with and without inhomogeneity embedded. A Monte Carlo dose calculation method is the most complex and accurate dose calculation method currently available and has a clear preference relative to other dose calculation methods in the quest for a dose delivery accuracy of 5% or better. Differences in dose calculation accuracy were observed for low density material with Monte Carlo slightly higher accuracy compared to advanced kernel methods.,, Monte Carlo dose calculation engines have the potential to meet, or even perform better than, the 3% (1σ) uncertainty requirement, regardless of beam geometry and patient composition. The level of error reduction in the dose calculation can be adjusted at the user level with the option Monte Carlo Variance (MCV) in Monaco. Decrease in MCV results in increase in calculation time and vice versa. To minimize the high frequency spatial fluctuations, MONACO permits smoothing of the fluence. The number of segments will be increased with decrease in Fluence Smoothing Factor (FSF) resulting in high calculation time. The purpose of this study is to scrutinize the influence of MCV with fluence smoothing for VMAT treatment planning
| » Methods and Materials|| |
The effects of MCV on quality of the treatment plan were evaluated on the basis of two aspects. First, the effects of MCV on the physical parameters such as MU per beam, number of segments, and total treatment delivery time. Second, the clinical parameters such as dose received by normal structures (OAR) and planning target volume (PTV). Homogeneity index and isodose coverage to the target volume was also evaluated. To analyze the variation of MCV on these parameters, five clinical cases of carcinoma esophagus were selected. The standard of care for esophagus is three dimensional conformal radiotherapy (3DCRT) followed by a 3DCRT or IMRT boost to primary lesion or to the postoperative bed. IMRT has not been found to give a better result in terms PTV dose homogeneity and lungs paring. For our study, carcinoma of the esophagus cases were planned for a conventional dose of 59.4 Gy in 33 fractions using single arc. The FSF and MCV were varied to generate different treatment plans. TPS under consideration allows changing of MCV between 1% and 10%. Increase in MCV will increase the simulation histories, leading to a higher time in dose calculation. The FSF influences the efficacy of MCV by increasing or decreasing the number of segments. Decrease of FSF increases total number segments and hence improved the dosimetric quality of the treatment plan and total delivery time. The increase in the total number of segments increases the dose simulation histories, which bring down the effective MCV (the effective MCV is the one that is achieved after the completion of dose calculation. For this study, the optimization parameters were kept fixed and 45 treatment plans were generated for all combinations of MCV of 1%, 3%, and 5% and FSF of 1, 3, and 5, with a fixed calculation grid of 0.3 mm. The minimum requirement for a plan to be valid for the study if the 95% of the PTV receives at least 95% of the prescribed dose. The dose homogeneity index (DHI) was used as a measure of PTV and is defined as the ratio between the doses to 95% (D95) and 5% (D5) of the volume of the PTV.
| » Results|| |
The result has been analyzed for physical and clinical parameters.
The number of segments, treatment time, and MU for different plans were tabulated in [Table 1]. It is evident from the table that the increase in FSF decreases the number of segments and hence in treatment time. However, decrease in segment number and treatment time did not have one-to-one correspondence or in other words a linear relationship. Out of three FSFs, five gives least of the segment number hence the treatment time; the variation in MCV not influencing any of the physical parameters. All the three plans, with 1%, 3%, 5% MCV, for a given FSF reads exactly the same number of segments with little different MUs and hence in treatment time. The difference in MU reads a maximum value of 19.3 MU [Table 1] for a given FSF.
|Table 1: Physical parameters for different Monte Carlo variance with fluence smoothing factor|
Click here to view
The clinical parameters have been analyzed for target and normal structures-heart, lung and spine.
The minimum requirement on PTV in each is to make it covet with 95% of the prescription to its 95% volume. However, it was not possible to achieve these constraints exactly, and there has been a small variation in the dose to PTV as this may not have a clinical relevance. Out of 45 plans, 68.9% were meeting the constraints of 95% of the prescription dose to 95% PTV. In the rest of the plans, 17.8% met with more than 94% PTV receiving 95% of the prescribed dose. In the worst scenario, 93.6% target was covered with 95% of the prescription (5643 cGy). MCV and FSF were varied to create other plans, which resulted in variations in PTV and OAR doses. The variation of FSF, for a constant MCV, results in maximum and minimum differences in target coverage by 3.6% and 0.6%, respectively. The variation of MCV, for constant FSF, had influenced a target coverage variation of maximum 2.0% and minimum 0.1%. The DHI for a particular patient for all nine plans were minimally affected (<.01; [Figure 1]). It was difficult to separate the DHI differences for MCV and FSF variations.
|Figure 1: Variation of planning target volume dose as a function of Monte Carlo variation (%) (left side) and as a function of fluence scaling factor (right side)|
Click here to view
The dose received by heart was evaluated for volume receiving 30 Gy. [Figure 2] indicates the observed variation MCV and FSF as a function of heart dose, for five study patients. [Figure 2] indicates the variation of FSF results in some substantial difference for three patients (Patients 2, 3, 5) with a maximum variation of 12.79% and not more than 2.2% for rest of two patients). In general, heart dose increases with increase in FSF for a particular MCV. The maximum heart dose was observed for FSF = 5 for each value of MCV tested. The lowest heart dose was observed for FSF = 1 except for first patient. The maximum and minimum heart dose variations were noted as 12.79% and 2.2% among the plans for variation of FSF.
|Figure 2: Variation of heart dose (V30-volume receiving 30 Gy) as a function of fluence smoothing factor (left side) and Monte Carlo variance|
Click here to view
The variation of MCV yielded a different result than when FSF was varied. The variation of heart dose was minimally affected with the variation of MCV tested for a fixed FSF. The maximum observed difference in the heart dose was 1.6%. The minimum value of heart dose variation was as low as 0.23%.
Lung and spine doses
The combined right and left lung volumes have been considered for the analysis. The spinal cord was evaluated for the dose received to 1% or 2 cc whichever was lesser of its volume. As stated previously, three out of five patients showed heart dose variations (Patient 2, 3, and 5). However, only patient 2 indicated a maximum lung dose variation of 4.3%. The remaining cases revealed less than 1% lung dose difference for variation of FSF when MCV taken as parameter. Similar result observed when MCV were varied [Table 2]. Patient 2 showed the maximum difference of 2.2%. Overall, 94% of plans resulted in lung dose difference less than 1% for varied MCV; 80% of plans for variation of the FSF. The variation in dose to spinal cord was not appreciably different for MCV and FSF variations. The highest standard deviation on spinal cord dose was 0.6 Gy.
|Table 2: Effect of Monte Carlo variance with fluence smoothing factor on combined lungs of five patients|
Click here to view
| » Discussion|| |
This study provides the results on varying either the MCV or the FSF independently. Increase in FSF results in reduction in the number of segments as shown in the [Table 1]. A good radiotherapy treatment plan can be considered as an optimized analytical solution for the dosimetric and physical parameter variables that lower dose to the critical structures, minimize hot and cold volumes, and lower treatment time (hence MU) while not compromising the prescribed dose coverage to the tumour volume, which is assumed to be a constant. So a set of good radiotherapy plan can be achieved in many ways with a little difference in these parameters and variables without a significant clinical impact. Our results based on the variation in MCV while keeping the FSF constant did not show a significant variation in the coverage of the PTV and normal tissue structure doses and physical parameters. However, this was not the case when FSF was varied, keeping MCV constant, that is, inferior treatmentplans resulted. It is interesting to note that the variation in heart dose was highest among all organs at risk for the variations in MCV and FSF tested. The maximum heart dose difference was 12.79% among the plans when FSF was varied keeping MCV constant. However, the same variation, yields to a significantly lesser maximum dose variation (4.3%) for lung. The number of plans exceeding the dose difference of 3% with the variation of FSF was noted as 20% and 53.4% for lung and heart, respectively; 26.6% plans are having heart dose difference more than 5% for FSF variation. The spinal cord has not shown more than 0.6 Gy. The heart dose was reaching its maximum value when the FSF was 5 for all MCVs. The variation of MCV had not given high (1.64%) heart dose variation for the plans with fixed FSF. Ninety percent of the plans showed less than 1% heart dose difference when 94% of the plans showed less than 1% lung dose difference (for constant MCV and varying FSF the maximum PTV dose variation observed to be 0.7% for the plans with different MCV. The Monte Carlo Variance and the Fluence Smoothing Factor variations were not influencing much in PTV coverage as it was in normal structures. The variation in FSF resulted in notable difference in dosimetric as well as physical parameters for the treatment plans, whereas the variation of the MCV does not. The MONACO user manual suggests to use less than or equal to 1% MCV variation for the quality assurance of the clinical plans for better accuracy. However, analyses of our clinical plans did not support this criterion. Ma et al. suggested that MCV less than 5% will give negligible variations among the plans, similar results were observed in this study. The variation on MCV shows a good dosimetric variation if lesser number simulation histories and single or lesser number of beams were considered. When a linear accelerator is modeled or simulated with a Monte Carlo technique, standard dose planes are considered, error in calculated and measured dose minimized for a particular MCV by reducing the noise level. These planes are accurate calculation planes with minimum noise, however, when the MCV is varied the noise level changes hence the calculation accuracy. Another important aspect of our study was to find the best possible MCV for MONACO. For a single field and less number of histories, the system gives the effective MCV near to the set MCV, for example, the set MCV of 5%, 4%, 3%, 2%, and 1% leads to an effective MCV of 5%, 4%, 3%, 2%, and 0.9%, respectively [Table 3]. It was seen that plans started with 1%, 3%, and 5% MCV showed an effective MCV of 0.2%, 0.5%, and 0.9%, respectively, on completion when higher number of beams as well as large history was the concern, that is, a typical clinical situation. In a clinical setting, the plans with 0.2%, 0.5%, and 0.9% MCV may not give good differences in the PTV and OAR doses. That may be the reason why we also did not find the differences in physical and clinical parameters with the variation in MCV. The optimization process also affects the variation in MCV. It's important to mention here, unlike for a fixed dose rate linear accelerator, lesser number of segments (physical parameters) does not necessarily leads to a lesser treatment time as there is a dose rate effect in case of VMAT plans. Using an MCV less than 1% may help in quality assurance but not in a clinical situation. The smaller MCV yields longer calculation time but similar results to utilizing 5% variance. The advantage of lower MCV (1-3%) has not been seen in esophagus cases used here possibly because of high degree of tissue heterogeneity. Whatever difference was measured in our test cases were because of variation in FSF. It is clear that in all cases the plan quality in terms of both dosimetric and physical quantities were maintained.
|Table 3: Dose measured and calculated at 10 cm for 10 ×10 field at 100 SAD for different Monte Carlo variances|
Click here to view
| » Conclusion|| |
Although XVMC is an open code system, the specific information such as linear accelerator details etc., for the of XVMC-based Monaco planning system is not available in public domain and is subject to copyright (personal communication with Elekta Medical System, Crawley, UK), thus it is not possible to check the behaviour of this planning system by changing actual code without the assistance of the vendor. So to evaluate the performance of this planning system under a clinical situation, two variables were considered (MCV and FSF). The effect of variation of MCV as well as FSF on the dosimetric and physical quality of the treatment plan was studied for a particular set of test cases (carcinoma of the esophagus) and the presented data are valid only for that site. The variation of MCV as well as FSF for other sites is subject matter of further investigation. It is revealed from this study that the MCV of 3% or less did not show any significant improvements in the plan quality (physical and clinical parameters) in comparison with MCV greater than 3%. The use of MCV between 3% and 5% gives similar results like 1% variance with lesser calculation time. Minimally detected differences in plan quality suggest that the optimum FSF can be set between 3 and 5.
| » References|| |
Otto K. Volumetric modulated arc therapy: IMRT in a single gantry arc. Med Phys 2008;35:310-17.
Crooks SM, Wu X, Takita C, Watzich M, Xing L. Aperture modulated arc therapy. Phys Med Biol 2003;48:1333-44.
Cameron C. Sweeping-window arc therapy: An implementation of rotational IMRT with automatic beam-weight calculation. Phys Med Biol 2005;50:4317-36.
Ulrich S, Nill S, Oelfke U. Development of an optimization concept for arc - Modulated cone beam therapy. Phys Med Biol 2007;52:4099-119.
Wang C, Luan S, Tang G, Chen DZ, Earl MA, Yu Cx. Arc-modulated radiation therapy (AMRT): A single-arc form of intensity-modulated arc therapy. Phys Med Biol 2008;53:6291-303.
Schreibmann E, Dhabaan A, Elder E, Fox T. Patient-specific quality assurance method for VMAT treatment delivery. Med Phys 2009;36:4530-5.
Nutting CM, Bedford JL, Cosgrove VP, Tait DM, Dearnaley DP, Webb S. A comparison of conformal and intensity-modulated techniques for oesophageal radiotherapy. Radiother Oncol 2001;61:157-63.
Bertelsen A, Hansen CR, Johansen J, Brink C. Single Arc Volumetric Modulated Arc Therapy of head and neck cancer. Radiother Oncol 2010;95:142-8.
Clemente S, Wu B, Sanguineti G, Fusco V, Ricchetti F, Wong J, et al
. SmartArc-based volumetric modulated arc therapy for oropharyngeal cancer: A dosimetric comparison with both intensity-modulated radiation therapy and helical tomotherapy. Int J Radiat Oncol Biol Phys 2011;80:1248-55.
Dobler B, Weidner K, Koelbl O. Application of volumetric modulated arc therapy (VMAT) in a dual-vendor environment. Radiat Oncol 2010;5:95.
Oelfke U and Scholz. Dose Calculation Algorithms. In: Schlegel W, Bortfeld T, Grosu AL, editors. New Technologies in Radiation Oncology Medical Oncology. Heidelberg: Springer; 2006. p. 187-96.
Ma CM, Pawlicki T, Jiang SB, Li JS, Deng J, Mok E, et al
. A L Monte Carlo verification of IMRT dose distributions from a commercial treatment planning optimization system. Phys Med Biol 2000;45:2483-95.
Chetty IJ, Curran B, Cygler JE, DeMarco JJ, Ezzell G, Faddegon BA, et al
. Report of the AAPM Task Group No. 105: Issues associated with clinical implementation of Monte Carlo based photon and electron external beam treatment planning. Med Phys 2007;34:4818-53.
Fipple M. Monte Carlo Dose Calculation for Treatment Planning. In: Schlegel W, Bortfeld T, Grosu AL, editors. New Technologies in Radiation Oncology Medical Radiology. Heidelberg: Springer; 2006. p. 197-206.
Krieger T, Sauer OA. Monte Carlo- versus pencil-beam-/collapsed-cone-dose calculation in a heterogeneous multi-layer phantom. Phys Med Biol 2005;50:859-65.
Fotina I, Winkler P, Künzler T, Reiterer J, Simmat I, Georg D. Advanced kernel methods vs. Monte Carlo-based dose calculation for high energy photon beams. Radiother Oncol 2009;93:645-53.
Reynaert N, van der S, Schaart D, vander Zee W, Tomsej M, vanvliet C, et al
. Monte Carlo Treatment Planning An Introduction. Report 16 of the Netherlands Commission on Radiation Dosimetry. 2006.
Marks LB, Yorke ED, Jackson A, Ten Haken RK, Constine LS, Eisbruch A, et al
. Use of normal tissue complication probability models in the clinic. Int J Radiat Oncol Biol Phys 2010;76:S10-9.
Ma CM, Li JS, Jiang SB, Pawlicki T, Xiong W, Qin LH, et al
. Effect of statistical uncertainties on Monte Carlo treatment planning. Phys Med Biol 2005;50:891-907.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3]
|This article has been cited by|
||Dosimetric impact of statistical uncertainty on Monte Carlo dose calculation algorithm in volumetric modulated arc therapy using Monaco TPS for three different clinical cases
| ||Mohandass Palanisamy,Khanna David,Manigandan Durai,Narendra Bhalla,Abhishek Puri |
| ||Reports of Practical Oncology & Radiotherapy. 2019; 24(2): 188 |
|[Pubmed] | [DOI]|