Htreatment plan verication

Erminia Infusino, Ph.D., AlessandGerardina Stimato, Ph.D., Luca BSara Ramella, M.D., Marcello BenDepartment of Radiotherapy, Universit Campus Bio-m

A R T I C L E I N F O

Article history:Received 2 January 2014Received in revised form6 March 2014Accepted 10 April 2014

Keywords:3DVH

the effectiveness oftype algorithm to

dical Dosimetrists.

RapidArc (RA) (Varian Medical Systems, CA) is a form of

use in 2008. VMAT delivery requires a stricter mechanical

nsity-modulatedof the simulta-

before treatment because of the complexity of the deliverybeam.1-4

journal homepage: w

Medical Do

Medical Dosimetry 39 (2014) 276281http://dx.doi.org/10.1016/j.meddos.2014.04.0040958-3947/Copyright 2014 American Association of Medical DosimetristsThe pretreatment dosimetric verication comprises a compar-ison of a measurement dose with the treatment planningsystem (TPS)-calculated dose. This can be performed either on a

Reprint requests to: Erminia Infusino, Department of Radiotherapy, UniversitCampus Bio-medico di Roma, via alvaro del portillo 200, Rome, Italy.

E-mail: e.infusino@unicampus.itvolumetric-modulated arc therapy (VMAT) that entered clinicalneous gantry speed, dose rate, and multileaf collimator apertureshape variations. Therefore, it requires a dosimetric vericationclinical importance of any disagreement between the measured and the calculated doto interpret; however, beam errors (either in delivery or in TPS calculation) can affectthe patient dose. Further research is needed to determinate the role of a PDP-accurately estimate patient dose effect.

& 2014 American Association of Me

Introduction performance for the linear accelerator than interadiation therapy (IMRT) delivery does becauserelevant differences increased with increasing QA passing rate, indicating that some of the largest dosedifferences occurred in the cases of high QA passing rates, which may be called false negatives. The

se is often difcultQuality assuranceArcCHECKGamma indexra Mameli, Ph.D., Roberto Conti, R.T., Diego Gaudino, Ph.D.,ellesi, Ph.D., Rolando Maria DAngelillo, M.D.,assi, Ph.D., and Lucio Trodella, M.D.

edico di Roma, Rome, Italy

A B S T R A C T

The purpose of this study was to perform delivery quality assurance with ArcCHECK and 3DVH system(Sun Nuclear, FL) and to evaluate the suitability of this system for volumetric-modulated arc therapy(VMAT) (RapidArc [RA]) verication. This software calculates the delivered dose distributions in patientsby perturbing the calculated dose using errors detected in uence or planar dose measurements. Thedevice is tested to correlate the gamma passing rate (%GP) and the composite dose predicted by 3DVHsoftware. A total of 28 patients with prostate cancer who were treated with RA were analyzed. RAtreatments were delivered to a diode array phantom (ArcCHECK), which was used to create a planneddose perturbation (PDP) le. The 3DVH analysis used the dose differences derived from comparing themeasured dose with the treatment planning system (TPS)-calculated doses to perturb the initial TPS-calculated dose. The 3DVH then overlays the resultant dose on the patient's structures using the resultantPDP beams. Measured dose distributions were compared with the calculated ones using the gammaindex (GI) method by applying the global (Van Dyk) normalization and acceptance criteria, i.e., 3%/3 mm.Paired differences tests were used to estimate statistical signicance of the differences between thecomposite dose calculated using 3DVH and %GP. Also, statistical correlation by means of logisticregression analysis has been analyzed. Dose-volume histogram (DVH) analysis for patient plans revealedsmall differences between treatment plan calculations and 3DVH results for organ at risk (OAR), whereasplanning target volume (PTV) of the measured plan was systematically higher than that predicted by theTPS. The t-test results between the planned and the estimated DVH values showed that mean valueswere incomparable (p o 0.05). The quality assurance (QA) gamma analysis 3%/3 mm showed that in allcases there were only weak-to-moderate correlations (Pearson r: 0.12 to 0.74). Moreover, clinicallyInitial experience of ArcCHECK and 3DV software for RapidArc

ww.meddos.org

simetry

set

E. Infusino et al. / Medical Dosimetry 39 (2014) 276281 277beam-by-beam basis using 2-dimensional (2D) measurement oron a whole-treatment basis using a 2D-measured dose plane or a3-dimensional (3D) measurement. In many cases, the lm isreplaced with a diode array or electronic portal imaging devicethat provides a beam-by-beam dose analysis rather than a com-posite dose analysis.

Conventional VMAT QA is usually performed by applying thepatient plan to a phantom with simple geometry and comparingthe measured and the calculated phantom dose distribution. Whencomparing the measured and the calculated dose in phantom,the gamma index (GI) that combines the percentage dose differ-ence (%DD) and the distance to agreement (DTA) is calculated foreach pixel. The gamma passing rate (%GP) (the portion of pixels

Fig. 1. Sagittal, axial, and coronal dose planes for a representative datathat has a GI less than 1) is then calculated for judgment on the QAresult.

There have been many studies on suggested acceptance levels forplanar IMRT-VMAT QA.5-10 Some of these studies base action levelson retrospective statistical analysis of the performance levels/metricsthat have been achieved over many plans and IMRT beams.5-10

In a report of the American Association of Physicists inMedicine Task Group 1195 and other studies5-9 as well, the 3%/3-mm criterion is common and is employed as the composite DTAmetric; therefore, the 3% dose difference and the 3-mm DTA

Fig. 2. Sample differences between the DVH obtained by TPS and the composite docriteria were reported as those most commonly used by clini-cians11 in per patient planar IMRT QA.

It was shown in a recent study that per-beam planar %GPs donot predict the clinical effect on the patient in terms of changes indose-volume histogram (DVH) values for the clinical target volume(CTV) and organs at risk (OARs).12

Indeed there have been several software systems that claim thecapability to estimate patient dose based on QA measurement,such as the COMPASS system (IBA-Wellhofer, Bahnhofstrasse590592 Schwarzenbruck, Germany), Dosimetry Check (Math Res-olutions, LLC, Gales Lane, Columbia), and 3DVH (Sun NuclearCorporation, Melbourne, Florida). The potential of using suchproducts to perform patient DVHbased IMRT QA is in need of

of the 28 data sets studied. (Color version of gure is available online.)further investigation.In this study, we used the software program 3DVH with the

ArcCHECK to investigate the correlation among %GP obtainedduring standard per-beam pretreatment QA tests.

The 3DVH software purports to be able to take beam measure-ments from a plan and reconstruct the full 3D dose distributionwithin the patient for comparison with the TPS calculations.

The 3DVH system was designed to incorporate the beam-by-beam phantom doses (measured and calculated) back into thepatient's images, structures, and TPS dose using planned dose

se predicted by the 3DVH software. (Color version of gure is available online.)

The quality assurance on RA treatment plans is performed by creating a

patient and patient DVH using conventional planar IMRT-VMAT QA data as inputs.

for PTVs.

Table 1Examples of the average %GP of patients evaluated

Patient Mean %GP

1 98.12 99.03 98.54 98.05 98.46 97.67 97.0

E. Infusino et al. / Medical Dosimetry 39 (2014) 276281278verication plan in the Eclipse TPS. This was achieved by copying the treatmentplan onto volumetric phantom devices. The predicted dose to the detector planeswas calculated with a dose grid resolution of 2.5 mm, it is sufcient to keep thedose error within 2%.15 This value was adopted for RA treatment plans.

Pretreatment verications were performed for all patient plans by acquiringplane dose distributions of each treatment eld. Measurements were taken usingthe diode array, routinely used in our institute, with absolute dose calibration, andthe software ArcCHECK.perturbation (PDP) for estimating the delivered patient dose andDVHs in 3D.13 The software uses the dose differences foundbetween the ArcCHECK measurement and the TPS dose calculationfor each beam and projects those differences back into the TPS 3Ddose calculation to obtain an estimate of the actual delivered 3Ddose distribution.

Methods

All patients with prostate cancer underwent radical radiotherapy up to 80 Gy tothe prostate and seminal vesicles with standard fractionation (2 Gy/die). Patientswere immobilized by customized devices with an empty rectum and a moderatelyfull bladder. Every day treatment was veried by cone-beam computed tomog-raphy scans. Gross tumor volume included the prostate and seminal vesicles(proximal thirds), CTV was dened as gross tumor volume 3-mm circumferentialmargin, and planning target volume (PTV) was dened as CTV 5-mm circum-ferential margin. The whole PTV received 95% to 105% of the prescribed dose. Therectum, bladder, penile bulb, and femoral heads were set as OARs, and doseconstraints have been dened according to the quantitative analyses of normaltissue effects in the clinic criteria.14

RA technique

Treatment planning was performed using photon beams of TrueBeam, withnominal energies of 6 MV. RA plans were optimized for 2 arcs (rotation of 3581, from1791 to 1811 clockwise (CW) and counterclockwise (CCW)), where multileaf collima-tor (maximum: 5 mm/deg and 2.5 cm/s), dose rate (maximum: 600 MU/min), andgantry speed (maximum: 72 s/turn, i.e., 5 deg/s) are optimized simultaneously toachieve the desired degree of modulation. All dose distributions were computed withthe anisotropic analytical algorithm implemented in the Eclipse planning systemwith a calculation grid resolution of 2.5 mm.

A total of 28 DMLC prostate RA treatment plans were exported as DICOMradiation therapy (RT) data sets from the planning system. These plans weredelivered to a 3D diode array (ArcCHECK, Sun Nuclear). Measured doses werecompared with the dose planes from the original plan, and dose differences wereused as input to 3DVH to calculate a delivered dose in the patient.

GI evaluationTable 2Evaluation of the %DD mean for 28 DVHs

Structure Parameter DD/fz mean (Gy) %DD

PTV Dmean 0.07 2.8PTV D95% 0.02 2.2Rectum Dmean 0.04 0.5Bladder Dmean 0.03 1.2Penile bulb Dmean 0.10 6Femoral head dx Dmean 0.01 0.6Femoral head sx Dmean 0.03 2.5

dx = right; fz = fraction; sx = left.The goal is to produce clinically relevant metrics to replace conventional metricsthat are limited in both sensitivity and specicity.13,18 The major function of thePDP algorithm is to use conventional per-beam planar dose QA methods to feed asophisticated 3D perturbation system that corrects the original 3D patient dose asgenerated by the TPS and outputs a 3D patient dose grid that has built into it themanifestation of any errors detected by the planar QA.20

The PDP method does not introduce new sources of variation or error that mayoccur with an independent 3D dose algorithm (i.e., variations that might not beerrors but just differences of the new algorithm vs the TPS algorithm). PDP altersdose only if and where dose differences are detected in dosimetry arraysystems.21,22 As an output, the 3DVH software compares the DVHs calculated byTPS with the ones obtained from perturbation, and it shows dose distributions,both alone and as comparison, in the sagittal, coronal, and axial planes ofpatients.23 Figure 1 shows samples of sagittal, axial, and coronal dose planes.Figure 2 illustrates a comparison between DVH obtained by TPS and the compositedose predicted by the 3DVH software.

Results

Planar dose distributions acquired during pretreatment veri-cations, RT plan, structure set, dose, and computed tomographyimages (the latter is optional) exported from TPS were loaded onthe software 3DVH. Overall, 2 kinds of comparisons were made,the rst was a 2D/3D gamma evaluation, and the second was aDVHs comparison.

DVHs provided by the 3DVH (delivered dose distribution frompretreatment verications) were compared with those calculatedby Eclipse (planned dose distribution) using the following dose-volume parameters: Dmean for all structures, and Dmean and D95ArcCHECK is a cylindrical water-equivalent phantom with a 3D array of 1386diode detectors, arranged in a spiral pattern, with 10-mm sensor spacing.

The per-beam IMRT QA dose distributions of each treatment plan wereanalyzed by employing the GI method16; by adopting the method proposed byZhen et al.,17 %GP was generated for each pair of planes using the global gammacalculation method and the acceptance criterion, 3%/3 mm.

The gamma analysis of patient treatment plans presented within this sectionwas done using only a single criterion (3%/3 mm) owing to the lack of relevance ofthe 2%/2-mm and the 1%/1-mm criteria. In the per-beam planar analysis, allpatients had average %GPs of more than 95% for the 3%/3-mm criterion.

Software

The 3DVH uses the PDP algorithm to estimate the dose to be delivered to the

Table 3The DVH parameters calculated by the TPS Eclipse and by the software 3DVH forPTVs and OARs and the relative p value

Structure Parameter %DD Mean dose difference (Gy)/fz p value

PTV Dmean 2.8 0.1 o 0.05PTV D95% 2.2 0.02 0.26Rectum Dmean 0.5 0.04 0.55Bladder Dmean 1.2 0.03 o 0.05Penile bulb Dmean 6 0.1 o 0.05Femoral head dx Dmean 0.6 0.003 o 0.05Femoral head sx Dmean 2.5 0.03 o 0.05

dx = right; fz = fraction; sx = left.Differences in the DVH parameters were calculated as

%DDD3DVHDTPS

DTPS

100

where D3DVH was the dose value shown by the software 3DVH, andDTPS was the dose value reported by TPS. This analysis wasperformed, for each patient, on PTVs and some OARs of theanatomically treated district (the rectum, bladder, penile bulb,femoral head dx [right], and femoral head sx [left]).

We calculated mean DVH values obtained using both 3DVH andTPS. To verify whether our procedures were affected by systematicerrors, we performed a t-test statistical analysis between the meanvalues.

E. Infusino et al. / Medical Dosimetry 39 (2014) 276281 279The t-test compares the actual difference between 2 means inrelation to the variation in the data (expressed as the standarddeviation of the difference between the means). The signicance level for a given hypothesis test is a value for which a p value less than orequal to is considered statistically signicant. These values corre-spond to the probability of observing such an extreme value bychance. Therefore, p o 0.05 was considered incomparable. Thesedata are analyzed to determine the systematic error,, which can be

Fig. 3. Correlation between the %GP calculated using the global 3%/3-mm criterion and tfemoral sx. (Color version of gure is available online.)dened as the mean of differences between D3DVH and DTPS for allorgans.

In addition, we determined the degree to which the DVH valuesare associated using the Pearson correlation coefcient (r). Thiscorrelation test is used to measure the strength of a linearassociation between 2 variables. The Pearson correlation coef-cient, r, can take a range of values from 1 to 1. A valueof 0 indicates that there is no association between the 2 variables.

he %DD for Dmean of the PTV, PTV 95%, rectum, bladder, penile bulb, femoral dx, and

0 o |r| o 0.3 is a weak correlation0.3 o |r| o0.7 is a moderate correlation|r| 4 0.7 is a strong correlation

E. Infusino et al. / Medical Dosimetry 39 (2014) 276281280GI passing rates of 3%/3 mm were evaluated for all the patients.Table 1 shows average %GP calculated for some patients. The %DDwas calculated for all patients and is reported in Table 2. Mean DVHvalues, t-test p values, and systematic error are illustrated in Table 3.

The t-test result between the planned and estimated DVHvalues showed that mean values were incomparable (p o 0.05),this indicates that there were systematic errors, whereas D95means for PTVs were comparable (p 4 0.05).

The number of correlations changed for different structures:Fig. 3 shows the correlation between %GP calculated using theglobal method and the %DD for PTV and OARs' dose values.

The QA gamma analysis 3%/3 mm showed that in nearly every case,there was only a weak-to-moderate correlation between the GP% andthe absolute %DDs (Pearson r: 0.12 to 0.74, Table 4), with the exceptionof the rectum (p 0.74). The rectum is a unique organ for which anonsignicant difference between the planned and estimated dosevalues (p 0.55) has been observed, whereas a statistically signicantdifference has been observed for the others organs.

Conclusions

The aim of this article was to evaluate the predictive meaningof the GI, in terms of correlation between the %GP obtainedduring standard pretreatment QA tests and the dose discrepancybetween planned DVH and patients' perturbed DVH. This articleevaluated whether the standard action levels used by most clinicsA value greater than 0 indicates a positive association, that is, asthe value of one variable increases, so does the value of the othervariable. A value less than 0 indicates a negative association, thatis, as the value of one variable increases, the value of the othervariable decreases. The magnitude of the correlation coefcientdetermines the strength of the correlation. Although there are nohard-and-fast rules for describing correlational strength, generally

Table 4Indexes of correlation between %GP and %DD for different structures

Structure Parameter r

PTV Dmean 0.57PTV D95% 0.12Rectum Dmean 0.74Bladder Dmean 0.38Penile bulb Dmean 0.69Femoral head dx Dmean 0.37Femoral head sx Dmean 0.36

dx = right; sx = left.(3%/3 mm with a 95% passing rate and global normalization) arejustied.

The 3DVH is a useful tool to calculate delivered dose distri-butions during patient-specic IMRT and VMAT quality assur-ance. The intended use is to incorporate the ArcCHECK (orelectronic portal imaging device or MapCHECK) measurementsand patient's DICOM RT structure set, dose, and plan les toreconstruct the 3D dose within the patient and to allow thecomputation and comparison of the DVH curves, planned anddelivered, for the anatomical structures that were of interestduring the planning of the patient's treatment, renderingphantom-based dose analysis obsolete. The process requires noextra time when compared with standard QA with theArcCHECK.

3. Bedford, J.L.; Warrington, A.P. Commissioning of volumetric modulated arc

therapy (VMAT). Int. J. Radiat. Oncol. Biol. Phys. 73(2):537e45; 2009;Otto, K. Volumetric modulated arc therapy: IMRT in a single gantry arc. Med.Phys. 35(1):3107; 2008.

4. Guckenberger, M.; Richter, A.; Krieger, T.; et al. Is a single arc sufcient involumetric-modulated arc therapy (VMAT) for complex-shaped target vol-umes? Radiother. Oncol. 93(2):259e65; 2009.

5. Both, S.; Alecu, I.M.; Stan, A.R.; et al. A study to establish reasonable actionlimits for patient-specic quality assurance in intensity-modulated radiationtherapy. J. Appl. Clin. Med. Phys. 2:18; 2007.

6. Basran, S.; Woo, M.K. An analysis of tolerance levels in IMRT quality assuranceprocedures. Med. Phys. 35(6):23007; 2008.

7. Howell, M.; Smith, I.P.; Jarrio, C.S. Establishing action levels for EPID-based QAfor IMRT. J. Appl. Clin. Med. Phys. 9(3):1625; 2008.

8. Ezzell, A.; Burmeister, J.W.; Dogan, N; et al. IMRT commissioning: Multipleinstitution planning and dosimetry comparisons, a report from AAPM TaskIn our study, we started by verifying the consistency of 3DVHwithour clinical VMAT plans. Differences between planned and measureddose were obtained when verifying real patient plans. For theselected DVH dose values, we found a signicant difference betweenmean doses calculated by TPS and 3DVH (Tables 2 and 3), especiallyPTV, the measured planwas systematically higher than that predictedby the TPS. However, their interpretation in clinical terms based on acommon gamma analysis is not unmistakable and remains unclear.

The results of correlation analysis showed that all r values werelow. This proved the weak correlation between the %GP and theabsolute %DDs.

Discussion

The 3DVH validity was demonstrated in previousarticles.8,10,12,20,21 In our study, we started by verifying the con-sistency of 3DVH with our clinical IMRT plans.

The patient dose QA methodology is used to accurately esti-mate the effect of errors on patient anatomy dose metrics.

There is a lack of correlation between conventional IMRT QAperformance metrics (%GPs) and dose differences in critical ana-tomical regions of interest. The most common acceptance criteriaand published actions levels therefore have insufcient, or at leastunproven, predictive power for per patient IMRT QA. However,these criteria does not ensure that clinically acceptable dose errors.

There have not yet been correlation studies to prove, ordisprove, whether these accepted methods for IMRT QA and theirassociated acceptance criteria are good predictors of clinicallyrelevant patient dose errors in per patient IMRT QA.

Recent experimental studies, which reveal the limited sensi-tivity of gamma analysis to patient dose deviation under differentIMRT errors, have been carried out.16,18,19,24 In fact, it was shown ina recent study that per-beam planar %GPs do not predict theclinical effect on the patient in terms of the changes in DVH valuesfor the CTV and OARs,12 which has questioned the feasibility of %GP-based IMRT QA.

The gamma analysis did not give clear information on whether aplan should be accepted. Kruse18 has recently shown that theaverage passing rate that would be clinically acceptable for onepatient could be unacceptable for another. Consequently, pretreat-ment checks should be done taking into account the clinicaltolerances of the OAR and the PTV and not relying on %GPs. To dothis, there is a need for a tool, such as 3DVH software, that couldshow the effect of the measured planar differences in the histogram.

References

1. Vanetti, E.; Clivio, A.; Nicolini, G.; et al. Volumetric modulated arc radiotherapyfor carcinomas of the oro-pharynx, hypo-pharynx and larynx: A treatmentplanning comparison with xed eld IMRT. Radiother. Oncol. 92:1117; 2009.

2. Verbakel, W.F.; Cuijpers, J.P.; Hoffmans, D.; et al. Volumetric intensity-modulated arc therapy vs. Conventional IMRT in head-and-neck cancer:A comparative planning and dosimetric study. Int. J. Radiat. Oncol. Biol. Phys.74:2529; 2009.Group 119. Med. Phys. 36(11):535973; 2009.

9. Pawlicki, T; Yoo, S; Court, L.E.; et al. Process control analysis of IMRT QA:Implications for clinical trials. Phys. Med. Biol. 53(18):5193205; 2008.

10. Van Dyk, J.; Barnett, R.B.; Cygler, J.E.; et al. Commissioning and qualityassurance of treatment planning computers. Int. J. Radiat. Oncol. Biol. Phys.26:26173; 1993.

11. Nelms, B.E.; Simon, J.A. A survey on planar IMRT QA analysis. J. Appl. Clin. Med.Phys. 8(3):7690; 2007.

12. Nelms, B.E.; Zhen, H.; Tome, W.A. Per-beam, planar IMRT QA passing ratesdo not predict clinically relevant patient dose errors. Med. Phys. 38:103744;2011.

13. Nelms, B.E.; Simon, W.E. U.S. patent 7,945,022 May 17, 2011.14. Marks, L.B.; Yorke, E.D.; Jackson, A.; et al. Use of normal tissue complication

probability models in the clinic. Int. J. Radiat. Oncol. Biol. Phys. 76:S109;2010.

15. Liu, T.; Rajaguru, P.; Dieck, G.; et al. SU-E-T-70: Comparison of two 3D gammaindex calculation schemes. Med. Phys. 38(6):3501; 2011.

16. Waghorn, B.J.; Meeks, S.L.; Langen, K.M. Analyzing the impact of intrafractionmotion: Correlation of different dose metrics with changes in target D95%.Med.Phys. 38:450511; 2011.

17. Nelms, B.E.; Zhen, H.; Tom, W.A. Conventional IMRT QA analysis metricsdo not predict clinically relevant patient dose errors. Med. Phys. 38:1037;2011.

18. Kruse, J.J. On the insensitivity of single eld planar dosimetry to IMRTaccuracies. Med. Phys. 37(6):251625; 2010.

19. Yan, G.; Liu, C.; Simon, T.A.; et al. On the sensitivity of patient-specic IMRT QAto MLC positioning errors. J. Appl. Clin. Med. Phys. 10:2915; 2009.

20. Zhen, H.; Nelms, B.E.; Tom, W.A. Moving from gamma passing rates to patientDVH-based QAmetrics in pretreatment dose QA.Med. Phys. 38(10):547789; 2011.

21. Olch, A.J. Evaluation of the accuracy of 3DVH software estimates of dose to virtualion chamber and lm in composite IMRT QA. Med. Phys. 39(1):816; 2012.

22. Both, S.; Alecu, I.M.; Stan, A.R.; et al. A study to establish reasonable actionlimits for patient specic quality assurance in intensity-modulated radiationtherapy. J. Appl. Clin. Med. Phys. 8(2):18; 2007.

23. Dempsey, J.; Romeijn, H.; Li, J.; et al. A Fourier analysis of the dose gridresolution required for accurate IMRT. Med. Phys. 32:3808; 2005.

24. Feygelman, V.; Zhang, G.; Stevens, C.; et al. Evaluation of a new dose QA device,or: The X's and O's of 3D dosimetry arrays. J. Appl. Clin. Med. Phys. 12:334668;2011.

E. Infusino et al. / Medical Dosimetry 39 (2014) 276281 281

Initial experience of ArcCHECK and 3DVH software for RapidArc treatment plan verificationIntroductionMethodsRA techniqueGI evaluationSoftware

ResultsConclusionsDiscussionReferences