ORIGINAL ARTICLE Year : 2016  Volume : 41  Issue : 4  Page : 214218 Clinical implementation of an electron monitor unit dosimetry system based on task group 71 report and a commercial calculation program Huijun Xu^{1}, Mariana Guerrero^{1}, Shifeng Chen^{1}, Xiaocheng Yang^{2}, Karl Prado^{1}, Colleen Schinkel^{3}, ^{1} Department of Radiation Oncology, University of Maryland School of Medicine, Baltimore, MD, USA ^{2} Radiation Oncology Center, Good Sam Hospital, MedStar RadAmerica, Baltimore, MD, USA ^{3} Department of Medical Physics, Tom Baker Cancer Centre, Calgary, AB, Canada Correspondence Address: Many clinics still use monitor unit (MU) calculations for electron treatment planning and/or quality assurance (QA). This work (1) investigates the clinical implementation of a dosimetry system including a modified American Association of Physicists in Medicinetask group71 (TG71)based electron MU calculation protocol (modified TG71 electron [mTG71E] and an independent commercial calculation program and (2) provides the practice recommendations for clinical usage. Following the recently published TG71 guidance, an organized mTG71E databook was developed to facilitate data access and subsequent MU computation according to our clinical need. A recently released commercial secondary calculation program  Mobius3D (version 1.5.1) Electron Quick Calc (EQC) (Mobius Medical System, LP, Houston, TX, USA), with inherent pencil beam algorithm and independent beam data, was used to corroborate the calculation results. For various setups, the calculation consistency and accuracy of mTG71E and EQC were validated by their crosscomparison and the ion chamber measurements in a solid water phantom. Our results show good agreement between mTG71E and EQC calculations, with average 2% difference. Both mTG71E and EQC calculations match with measurements within 3%. In general, these differences increase with decreased cutout size, increased extended source to surface distance, and lower energy. It is feasible to use TG71 and Mobius3D clinically as primary and secondary electron MU calculations or vice versa. We recommend a practice that only requires patientspecific measurements in rare cases when mTG71E and EQC calculations differ by 5% or more.
Introduction In many clinics, the electron treatment planning is still performed by "forcing" pointdose monitor unit (MU) calculations onto treatment planning system (TPS). TPS is only used to generate a relative dose distribution on the patient CT images, although the more complicated algorithms such as Gaussian pencil beam [1],[2] and Monte Carlo [3],[4] were released. The electron dose per MU is not always easy to predict due to the variation in electron scattering between different models of linear accelerators and treatment setups. [5],[6],[7],[8] In 2014, the American Association of Physicists in Medicine published task group (TG) report No. 71. [9] With inclusion of the electron dosimetry considerations mentioned in other TG reports, [10],[11],[12] TG 71 explicitly proposed a unified definition for the dosimetric quantities in electron MU calculations and provided instructions to determine these quantities. It is beneficial to have such a wellestablished electron MU calculation protocol implemented in clinic. In keeping with common QA practices discussed in TG71, [9] it is essential to utilize an appropriate secondary calculation method. A recently released commercial secondary calculation program  Electron Quick Calc (EQC) module of a QA software Mobius3D (Mobius Medical System, LP, USA)  can be used for this purpose. The EQC utilizes the pencil beam redefinition algorithm based on two studies [13],[14] and airgap correction technique for extended source to surface distances (SSDs). Its different calculation engine and source of beam data make EQC a completely independent doublecheck of our modified TG71 electron (mTG71E) protocol. This work is the first report on the clinical implementation of TG71 and EQC for electron dosimetry for Varian Clinac machines. This work describes two dose calculation methodologies and compares the calculation results with each other and with validation measurement. By providing a concise commissioning procedure and the benchmark data, this work intends to give readers a better understanding of the potential differences and limitations of using these two electron MU calculations in clinic. Methods Modified task group71 electron protocol design and dosimetry data measurements A new databook was developed to utilize an mTG71E protocol as a primary electron MU calculation system for our Varian Clinac Series linear accelerators. The electron data in the databook were acquired based on the measurements on a Varian Clinac iX machine for electron energies 6, 9, 12, 16, and 20 MeV. A set of square Cerrobend cutouts was designed and crafted for each applicator (6 cm × 6 cm, 10 cm × 10 cm, 15 cm × 15 cm, 20 cm × 20 cm, or 25 cm × 25 cm). The cutouts were made in the following square field sizes (at isocenter): 2, 3, 4, 6, 10, 15 and 20 cm up to the size of the applicator. All measurements were made using a Wellhofer CC04 ion chamber (IBA, Nashville, TN, USA) (4 mm inner diameter) in a large (50 cm × 50 cm × 50 cm) water tank. The mTG71E was evolved from TG71 to adapt to our old data and measurement need, and we expected the differences between TG71 and mTG71E in MU calculation to be small for most clinical cases. Percentage depth doses (PDDs) data using various cutouts on a 25 cm × 25 cm applicator collected at 100 cm SSD were selected as representative data for mTG71 calculations. As to the influence of applicator size on PDD, parameters including R 90 , R 80 , R 50 , and R p and the Xray tail for 25 cm × 25 cm and 10 cm × 10 cm applicator were compared. Beyond d max , most PDDs agreed within 1 mm while the most noticeable difference was in the Xray tail and the differences were <1.5%. The dependence of PDD on SSD for all electron energies was checked by comparing the 15 cm × 15 cm open applicator PDDs for three different SSDs: 100, 105, and 110 cm. The difference was found to be <1 mm. In our practice of mTG71E, the electron output factor in the TG71 report is separated into the applicator factor and the cutout factor. The applicator factor for each energy is the dose per MU of an open applicator field to the open reference applicator field (which is 15 cm × 15 cm in our clinic), both at the same d max for the open reference applicator field since the correction of dosimetric effect of electron measurement at different depth is <0.1%. The cutout factor is the dose per MU of a cutout field to the open applicator field. These two dose readings were measured at d max for the cutout and the d max for the applicator, respectively. Due to the different depth, corrections of watertoair stopping power ratios and replacement factor were applied based on TG70 report [11] and Zhang et al. [15] To get air gap factors f air , the effective SSD was first calculated from the cutout measurements at different SSDs, following the TG71 formalism. While TG71 assumes f air and applicator size to be independent, our air gap factors include this dependency. Commercial dose calculation program (Mobius EQC) In this study, Mobius EQC (version 1.5.1) was used as a secondary dose calculation program. The beam data used in EQC follow the description in TG71 report. Specifically, reference applicator open field in EQC is 10 cm × 10 cm applicator open field; output factor is directly used; air gap factor is assumed independent of applicator size. These differences were considered when comparing beam data. The mTG71 and EQC calculations use dosimetric quantities for various depths, field sizes (from 2 cm to applicator size) and SSDs (99110 cm for mTG71 and 98115 cm for EQC). The calculation results using mTG71E and EQC were compared for electron beams of energy 6, 9, 12, 16, and 20 MeV using various cutout sizes and SSDs of 100, 105, or 110 cm. Validation measurements Validation measurements were performed for electron beams of energy 6, 9, and 16 MeV, using a CC04 ion chamber (Scanditronix Wellhofer, IBA Dosimetry, Bartlett, TN, USA) in a solid water phantom (Plastic Water, Best Industries, Nashville, TN, USA). Five customized cutouts of varying rectangular sizes defined at an SSD of 100 cm  6 cm × 12 cm, 4 cm × 12 cm, 6 cm × 8 cm, 4 cm × 8 cm, 3 cm × 6 cm  were crafted. The irregular cutouts were not included in this work due to their variability and that their field sizes can be estimated using rectangular field sizes. [9] The criteria for the selection of these field sizes included approximations to clinical scenarios and tests of the limitations of the algorithm in small dimension and varying aspect ratio situations. Cerrobend cutouts were designed to fit in either a 10 cm × 10 cm or a 15 cm × 15 cm applicator. The measurement depth was at a depth of maximum dose for each cutout field; SSD was either 100 cm or 110 cm. The measured ionization per MU was converted to dose per MU after correcting stopping power ratios and P repl at different depths. The dose per MU was then normalized to that at reference setup to eliminate daytoday machine output fluctuation. Results Modified task group71 electron versus Mobius EQC calculations The mTG71E and EQC MU calculations for the square and rectangular cutouts are illustrated in [Figure 1] for 9 MeV and 16 MeV, respectively. Overall, the mTG71E and EQC calculations give reasonably consistent results. The difference between two calculations is 0%5%, and in most cases, it decreases with smaller SSD and/or increased cutout size. The results also show that the EQC MU is often systematically higher than mTG71E because of different cutout factors and air gap factors used in two calculations.{Figure 1} Calculations versus chamber measurement To validate the calculation results for mTG71E and Mobius EQC, measurement using a CC04 ion chamber was performed in solid water for 9 MeV and 16 MeV electron beams. The dose ratios between mTG71E calculation and measurement and between EQC and measurement are plotted in [Figure 2]. Both mTG71E and EQC showed good consistency with chamber measurements: the mTG71E measurement ratio varies within 2.3%, and the EQCmeasurement ratio is within 1.5%.{Figure 2} [Figure 2]a and [Figure 1] indicate that deviations between mTG71E and solid water measurement and between mTG71E and EQC often increase for lower energy, smaller cutout field, and larger SSD. To investigate the differences at the lowest energy, we did a chamber measurement for 6 MeV, with a 3 cm × 6 cm cutout at 110 cm SSD, and compared to mTG71E and EQC calculations. Although mTG71EEQC difference is 5.3% in this case, both calculations differ from the measured value by <3%. Discussion This work focuses on the practical clinical implementation of a dose calculation system composed of two algorithms  the mTG71E and the Mobius EQC module  which were recently published/released. The Mobius EQC uses an independent algorithm and dataset and is becoming more frequently used in radiation oncology. To the best of our knowledge, there is no report/publication to verify this independent algorithm and its accompanying dataset. Through the discussion of the limitation of these two algorithms and setting the benchmark for their clinical implementation, this work also recommends the tolerance and action levels of different clinical scenarios with different calculation discrepancies of calculations. This work will be significant and useful for clinics using or planning on using TG71based electron dosimetry and/or Mobius software. It is feasible to use a TG71based MU calculation and the EQC program as primary and secondary electron calculations or vice versa. In general, the mTG71E and EQC calculations give consistent results with each other (05%) and with the chamber measurements in solid water (<2.5%) for different energies, field sizes, and SSDs as shown in [Figure 1] and [Figure 2]. The calculation difference between mTG71E and EQC mainly resides in different cutout factors and air gap factor. We found these dosimetric quantities between mTG71E and EQC all agree within 3%, except those for the 2 cm × 2 cm cutout field at 110 cm SSD and lower electron energies (i.e., 6 MeV and 9 MeV). The cutout factor difference sometimes cancels out the airgap factor difference. In most cases, we expect that the cumulative effect of dosimetric quantity difference between mTG71E and EQC on MU calculation is within 5%. When the scenarios involve low energy, a small cutout, and large extended SSD, we could expect a larger discrepancy between the mTG71E and EQC results. We recommend 5% as the difference criterion between the two calculations since the inherent beam data and algorithm differences may lead to large discrepancies in unusual clinical circumstances (small fields, large SSDs, etc.). Our practice in using mTG71E and EQC calculations is summarized as follows: if the difference between the two is within 3%, the MU calculated with mTG71E is used for electron treatment planning; if the difference lies between 3% and 5%, a physicist is consulted and if the difference exceeds 5%, a patientspecific measurement will be performed and the physicist will determine the MU setting to be used. This study investigates the performance of mTG71E and EQC for Varian Clinac accelerators exclusively. The dataset of other linear accelerator manufacturers (i.e., Elekta and Siemens [16],[17] or of different design (e.g., Varian TrueBeam [18],[19],[20] may be quite different from what we use in this work. [21] However, the same benchmark methodology can be used for the different machines. Therefore, this work is still useful to extend the applications of mTG71E and EQC to other machines. Conclusion The modified TG71 method and the Mobius EQC program have been implemented as an initial MU calculation and a redundant check for electron dosimetry in our clinic. Although they use independent algorithms and beam data, their dosimetric quantities for calculation are similar and their calculation results for various electron energies and setup are both reasonable. As these two calculation methods agree with each other and match measurement results to an acceptable level, they represent a reliable system to use for electron dosimetry. We propose a 5% agreement tolerance between both systems and recommend a patientspecific measurement for cases in which differences exceed this criterion. Acknowledgments We would like to thank Dr. Katja Langen and Dr. Sina Mossahebi for their help in this work. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. References


