- Email: [email protected]
Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors
Physica Medica xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Physica Medica journal homepage: http://www.physicamedica.com
Original paper
Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors Sankar Arumugam a,b,⇑, Aitang Xing a,b, Tony Young a,c, David Thwaites a,c, Lois Holloway a,b,c,d a
Liverpool and Macarthur Cancer Therapy Centres and Ingham Institute, Liverpool, New South Wales, Australia University of New South Wales, Sydney, New South Wales, Australia c Institute of Medical Physics, School of Physics, University of Sydney, Sydney, New South Wales, Australia d Centre for Medical Radiation Physics, University of Wollongong, Wollongong, New South Wales, Australia b
a r t i c l e
i n f o
Article history: Received 31 May 2016 Received in Revised form 23 August 2016 Accepted 22 September 2016 Available online xxxx Keywords: VMAT Dosimetric systems Delivery error
a b s t r a c t Aim: To study the sensitivity of three commercial dosimetric systems, Delta4, Multicube and Octavius4D, in detecting Volumetric Modulated Arc Therapy (VMAT) delivery errors. Methods: Fourteen prostate and head and neck (H&N) VMAT plans were considered for this study. Three types of errors were introduced into the original plans: gantry angle independent and dependent MLC errors, and gantry angle dependent dose errors. The dose matrix measured by each detector system for the no-error and error introduced delivery were compared with the reference Treatment Planning System (TPS) calculated dose matrix for no-error plans using gamma (c) analysis with 2%/2 mm tolerance criteria. The ability of the detector system in identifying the minimum error in each scenario was assessed by analysing the gamma pass rates of no error delivery and error delivery using a Wilcoxon signed-rank test. The relative sensitivity of the system was assessed by determining the slope of the gamma pass line for studied error magnitude in each error scenario. Results: In the gantry angle independent and dependent MLC error scenario the Delta4, Multicube and Octavius4D systems detected a minimum 2 mm error. In the gantry angle dependent dose error scenario all studied systems detected a minimum 3% and 2% error in prostate and H&N plans respectively. In the studied detector systems Multicube showed relatively less sensitivity to the errors in the majority of error scenarios. Conclusion: The studied systems identified the same magnitude of minimum errors in all considered error scenarios. Ó 2016 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.
1. Introduction Volumetric Modulated Arc Therapy (VMAT) generates highly conformal radiotherapy plans with reduced treatment delivery time [1]. These advantages come with added complexity at planning and delivery stages of the treatment. The plan specific dosimetric quality assurance (QA) check is an integral part of the treatment delivery chain in these techniques to ensure the accuracy of dose delivery [2]. The rotational delivery aspect of these treatment techniques demands dosimetric systems that are insensitive to beam incident angle and provide delivered dose information in multi dimensions with high spatial resolution.
⇑ Corresponding author at: Department of Radiation Oncology, Cancer Therapy Centre, Liverpool Hospital, Sydney, NSW 2170, Australia. E-mail address: [email protected] (S. Arumugam).
An ideal dosimetric system for the verification of advanced treatment techniques would provide high resolution 3D spatial information of the delivered dose. Gel dosimeters are well-suited for this purpose [3,4], however their widespread use in routine clinical practice is limited due to various factors such as the requirement of a high level of expertise, the requirement of expensive readout systems and high sensitivity of gel dosimeters to gel preparation, composition and ambient conditions. Currently available electronic dosimetric systems provide a reconstructed measured dose matrix for rotational treatment delivery by accounting for appropriate corrections to the beam incident angle on the detector elements available [5–7]. The detector element arrangements in modern electronic dosimeters vary widely; the systems have configurations with stationary and rotating 2D planar arrays, biplanar 2D arrays and helical grid arrays in an annular plane of a cylindrical phantom [5,6,8]. The measured dose matrix dimension varies from a two
http://dx.doi.org/10.1016/j.ejmp.2016.09.016 1120-1797/Ó 2016 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
2
S. Arumugam et al. / Physica Medica xxx (2016) xxx–xxx
dimensional matrix that measures single planar dose to multiplanar and complete three dimensional matrices depending on both the geometry of the detector arrangement and data processing used. Previous studies have investigated intrinsic characteristics, utilisation and limitations of modern detector systems in the dosimetric QA of advanced radiotherapy techniques [5,6,8–10]. Fredh, et al. [11] investigated four commercial dosimetric systems and their response to MLC bank position and dosimetric and collimator errors introduced to VMAT plans. Fredh et al. concluded that the sensitivity of each particular detector system’s ability to detect errors is specific to a given radiotherapy plan. Their study didn’t assess the relative sensitivity of the detector systems in identifying the errors. The purpose of this work is to study the ability of three commercial detector systems, Delta4, Multicube and Octavius 4D, in detecting delivery errors that are characteristic to VMAT. The clinical significance of minimum error detected by each dosimetric system was assessed by comparing the previously published dosimetric impact of the errors calculated in the Treatment Planning System (TPS) for the same data set by our research group [12].
2.3. Studied errors Three types of errors were intentionally introduced to the original plans considered in this study. In-house developed software was used to introduce the errors [12,15]. 1. Gantry angle independent MLC shift to simulate the systematic offset in the MLC position (ranging from ±1 mm to ±5 mm). 2. Gantry angle dependent MLC error (amplitude ranging from 1 mm to 5 mm), to simulate the sag in MLC position as a function of gantry angle [12]. 3. Gantry angle dependent dose error (amplitude ranging from 1% to 10%) to simulate the output fluctuation as a function of gantry angle [12]. The total monitor units (MU) of the VMAT arcs were maintained with the specified gantry angle dependent errors introduced to the relative weights of the control points (CPs). The error introduced plans were recalculated in the TPS and the treatment delivery files of these plans were exported to the LA for delivery.
2. Materials and methods
2.4. Clinical significance of introduced errors
2.1. Planning and delivery
The impact of introduced errors on the dose to the planning target volume (PTV) and organs at risk (OARs) was published in our previous work [12]. For the convenience of correlating the error detected by the dosimetric systems considered in this study the results are compared here. For prostate plans the impact of the studied errors on the dose delivered to 95% of PTV (D95), volume receiving 60 Gy (V60) for rectum and dose delivered to 50% (D50) of bladder was assessed. Similarly in H&N plans D95 to high dose PTV (HD-PTV), D95 to low dose PTV (LD-PTV) and the maximum dose delivered to a 2 cc volume (D2cc) of the spinal cord were considered. The percentage change in studied dose metrics with respect to the no-error plans was quantified to study the impact of errors.
The VMAT plans were generated using the SmartArc optimisation tool available in the Pinnacle v9.8 (Philips Healthcare, Fitchburg, WI, USA) Treatment Planning System (TPS). A 6 MV photon beam model for an Elekta-Synergy (Elekta Ltd, Crawley, UK) linear accelerator (LA) was used for planning. The details of the SmartArc optimisation algorithm used in Pinnacle can be found elsewhere [13]. Dose calculations for all plans in the study were performed using the Adaptive Convolution dose calculation algorithm available in the Pinnacle TPS with 2.5 2.5 2.5 mm3 voxel size. The structures representing the Elekta Evo table top with appropriate densities were included in the dose calculation in all patient and verification plans. The Synergy accelerator used in this study was equipped with an MLCi head and desktop pro, v 7.0, LA control software. Mosaiq, v 2.30.0D1, (Impac Medical Systems, Inc. California, USA) record and verification (R&V) system was used for the transfer of VMAT plans from TPS to LA.
2.2. VMAT plans Fourteen clinical VMAT plans (Seven each of prostate and head and neck (H&N)) were considered for the study. Some of the key characteristics of the plans are shown in Table 1. To assess the complexity of the plans considered in the study a multiplicative combination of leaf travel and modulation complexity score [14] (LT-MCS) was calculated using in-house software [14]. The mean (r) LT-MCS value for simple arc, prostate and H&N plans are shown in Table 1. The lower LT-MCS value represents higher complexity in the plan. As seen in Table 1 H&N plans were highly complex for the studied plan cohorts.
Table 1 Some of the important characteristics of the VMAT plans considered. Parameter
Arc length Coll angle Mean (r)LT-MSCv
Parameter value Prostate plans
H&N plans
240°–352° 10° 0.4(0.035)
352° 10° 0.07(0.05)
2.5. Studied dosimetric systems The following three commercial dosimetric systems were assessed for their ability to detect the errors considered in each error scenario. Table 2 shows some of the key characteristics of these systems. 2.5.1. Delta4 The Delta4 (ScandiDos AB, Uppsala, Sweden) is a cylindrical acrylic phantom containing 1069 diode detectors arranged in two orthogonal planes. The measurement planes pass through the centre of the phantom and are 45° oblique to the sagittal and coronal planes. Table 2 summarises some of the key characteristics of the Delta4 system and a more detailed description can be found elsewhere [5]. The measured signal is corrected for background, beam incident angle and ambient temperature. The manufacturer recommended dose calibration procedure was followed to correct for the detector array relative response and dose calibration. The verification treatment plans were recalculated on the pseudo CT data set representing the Delta4 geometry. All the dose calculations were performed with a density of 1.19 g/cm3 forced to Delta4 geometry in the TPS. ScandiDos Delta4 software (ScandiDos AB, Uppsala, Sweden) was used for the measurement and analysis of agreement to TPS calculated dose. 2.5.2. MatrixxEvolution with MultiCube MatrixxEvolution (IBA Dosimetry GmbH, Germany) is an ionchamber 2D array consisting of 1020 cylindrical shaped
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
3
S. Arumugam et al. / Physica Medica xxx (2016) xxx–xxx Table 2 The key physical characteristics and measured dose matrix of the dosimetric systems considered in the study. Detector systems characteristics
Number of detector elements and their type and arrangement Detector size Detector pitch Detector inactive region Final dose matrix
Detector system Delta4
Multicube
Octavius4D
1069 diode detectors Orthogonal planes 0.78 mm2 5 mm at central 6 6 cm2 and 10 mm outer 20 20 cm2 5 mm at central 6 6 cm2 and 10 mm outer 20 20 cm2 Orthogonal planes (with the option of 3D dose matrix)
1020 ion chambers Coronal plane 4.5 mm/ and 5 mm height 7.62 mm 3.1 mm Coronal plane
729 ion chambers 2D array in rotating phantom 5 mm 5 mm 5 mm 10 mm 5 mm 3D dose matrix
ion-chambers arranged in a 2D matrix with a 32 32 grid (Table 2). The MultiCube (IBA Dosimetry GmbH, Germany) is a cubical phantom of 30 cm 30 cm 25 cm dimension and can house the MatrixxEvolution which enables a planar dose measurement in the coronal plane. A detailed description of the MatrixxEvolution and its application in the quality assurance of rotational treatment with MultiCube can be found elsewhere [7]. The measured signal is corrected for background, angular response of the ion chambers and ambient temperature and pressure changes from the calibration measurement. The manufacturer provided calibration file was used to correct the response differences of the ion chambers. The absolute calibration of the device was performed as per the manufacturer’s recommendation. The verification treatment plans were recalculated on the CT data set of the MultiCube phantom with the MatrixxEvolution. The planar dose corresponding to the measurement plane of the MatrixxEvolution was exported from the TPS for comparison with measurement. The Omnipro I’mRT (v 6.1) software (IBA Dosimetry GmbH, Germany) was used to perform the measurements and comparison of measured and TPS dose matrices. 2.5.3. Octavius 4D 729 Octavius 4D 729 (PTW Freiburg GmbH, Germany) consists of a PTW 729 ion chamber 2D array housed in an Octavius 4D phantom. The PTW 729 detector array consists of 729 cubical ion chambers arranged in 2D matrix with a 27 27 grid. The Octavius4D phantom is a cylindrical phantom of 32 cm diameter and 34.3 cm length. The phantom rotates in synchrony with the gantry of the LA so that the measurement plane of the 2D array is always perpendicular to the beam axis. The gantry angle sorted 2D dose matrices are used for the reconstruction of 3D matrices with Percentage Depth Dose (PDD) data of the treatment beam measured at 85 cm SSD. The cumulative 3D dose matrix of the measured plan is calculated by adding all the reconstructed dose matrices. More details on the description and function of the Octavius 4D phantom and the PTW 729 array detector can be found elsewhere [16]. The signal measured by the PTW 729 array detector is corrected for background and ambient temperature and pressure differences from calibration measurements. The manufacturer provided calibration file was used to correct the response differences of the ion chambers in the 2D array. The absolute calibration of the device was performed as per the manufacturer’s recommendation. The verification treatment plans were recalculated on the pseudo CT data set representing a cylinder of dimensions equivalent to the Octavius4D phantom. The dose calculations were performed by forcing the density of 1.05 g/cm3 for the phantom. The Verisoft (v5.1) software (PTW Freiburg GmbH, Germany) was used to perform the measurements and 3D dose matrices were reconstructed with the voxel size of 2.5 mm 2.5 mm 2.5 mm. The comparison of TPS and reconstructed dose matrices was performed in Verisoft software. 2.6. Dose matrix comparison The measured and TPS calculated dose matrices were compared using c analysis [17] with 2% global dose tolerance and 2 mm
distance-to-agreement (DTA) tolerance criteria (2%G/2 mm). The maximum dose in TPS dose matrices of respective detector geometry was considered. The verification software included in the relevant dosimetric systems was used for the analysis of TPS and respective measured dose matrices. The dose matrix elements that receive less than 10% of the mean dose to PTV in respective detector geometry were not considered in the analysis. 2.7. Data analysis The minimum error detected by each of the dosimetric systems in prostate and H&N plans was determined by performing a Wilcoxon signed-rank test [18] between paired data of c pass rate of no error and error plans. The R statistical computing and graphics software was used for the calculation of the Wilcoxon test statistic (W) and p value [19]. In the Wilcoxon signed-rank analysis of data the minimum magnitude of a particular error was considered detected by a dosimetric system if the test statistic suggested a statistical difference (p < 0.05) between the paired no error and error pass rates. 2.8. Relative sensitivity of the systems to delivery errors The sensitivity of each dosimetric system to delivery errors was studied by determining the slope of the line (S) describing the gamma pass rate for no error and the studied range of errors in each error scenario. For the purpose of calculating slope, the gamma pass rate of each system for the studied error range was fitted using a linear least square fit. 3. Results 3.1. Impact of introduced errors on dose-volume metrics Figs. 1 and 2 shows the mean (r)% change in dose-volume metrics for the studied error scenarios in the prostate and H&N plans respectively. In both prostate and H&N plans the MLC errors resulted in the decrease in PTV D95 and the magnitude of decrease was proportional to the magnitude of error. In the gantry angle dependent dose error scenario the PTV D95 in prostate plans was not affected by the error whereas a small change was observed in H&N plans. 3.2. Change in gamma pass rate with errors The mean (r) c pass rate for no error and error introduced plans by each of the detector systems for prostate and H&N plans are shown in Figs. 3–5 for gantry angle independent MLC error, gantry angle dependent MLC error and dose error scenarios respectively. In both prostate and H&N plans the pass rate of all the detectors decreased, compared to no error plans, as the magnitude of MLC error increased in both gantry angle dependent and independent MLC error scenarios. In the gantry angle dependent dose error sce-
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
4
S. Arumugam et al. / Physica Medica xxx (2016) xxx–xxx
PTV D95 Bladder D50 Rectum V60
10.0 5.0 0.0 -5.0
-10.0 -15.0
No error 1 2 3 4 5 7 10
No error 1 2 3 4 5
No error ±1 ±2 ±3 ±4 ±5
No error 1 2 3 4 5 7 10
-25.0
No error 1 2 3 4 5
-20.0 No error ±1 ±2 ±3 ±4 ±5
% difference in DVH metric
15.0
Gantry angle Gantry angle Gantry angle dependent Gantry angle Gantry angle Gantry angle independent MLC dependent MLC dose error (%) independent MLC dependent MLC dependent dose error error (mm) error (mm) error (mm) error (mm) (%) PTV
OARs
Fig. 1. The mean (r) percentage change in PTV D95, Rectum V60 and Bladder D50 for the studied error scenarios in Prostate VMAT plans.
LD-PTV D95 HD-PTV D95
10.0
Spinal cord D2cc 5.0 0.0
No error 1 2 3 4 5 7 10
No error 1 2 3 4 5
No error ±1 ±2 ±3 ±4 ±5
No error 1 2 3 4 5 7 10
-10.0
No error 1 2 3 4 5
-5.0
No error ±1 ±2 ±3 ±4 ±5
% difference in DVH metric
15.0
Gantry angle Gantry angle Gantry angle dependent Gantry angle Gantry angle Gantry angle independent MLC dependent MLC dose error (%) independent MLC dependent MLC dependent dose error error (mm) error (mm) error (mm) error (mm) (%) PTV
OAR
Fig. 2. The mean (r) percentage change in D95 of High dose PTV (HD-PTV), Low dose PTV (LD-PTV), and Spinal cord D2cc for the studied error scenarios in Head and Neck VMAT plans.
100
90 80 70 60
Delta4 Multicube
90.0
Octavius4D pass rate (%)
pass rate (%)
100.0
Delta4 Multicube Octavius4D
80.0 70.0 60.0
50
50.0 40
No ±1 error
±2
±3 Prostate
±4
±5
No ±1 error
±2
±3
±4
±5
H&N
Treatment site and MLC error (mm) Fig. 3. The mean (r) gamma pass rates for no error and error introduced delivery in the gantry angle independent MLC error scenario for prostate and H&N plans.
40.0 No 1 error
2
3 Prostate
4
5
No 1 error
2
3
4
5
H&N
Treatment site and MLC error (mm) Fig. 4. The mean (r) gamma pass rates for no error and error introduced delivery in the gantry angle dependent MLC error scenario for prostate and H&N plans.
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
5
S. Arumugam et al. / Physica Medica xxx (2016) xxx–xxx
Delta4
100.0
Multicube
pass rate (%)
90.0
Octavius4D
80.0
mum 1 mm error in both MLC error scenarios in prostate plans and detected 1 mm and 2 mm errors in gantry angle independent and dependent MLC error scenarios in H&N plans. All three systems detected 3% gantry angle dependent dose errors in prostate plans and 2% error in H&N plans.
70.0 60.0
3.4. Relative sensitivity of the dosimetric systems in detecting errors
50.0 40.0
Prostate
10
7
No error 1 2 3 4 5
10
7
No error 1 2 3 4 5
30.0
H&N
Treatment site and dose error (%) Fig. 5. The mean (r) gamma pass rates for no error and error introduced delivery in the gantry angle dependent dose error scenario for prostate and H&N plans.
nario the c pass rate of the Multicube system showed minimum change (only 4% reduction from no error mean pass rate for 10% dose error) compared to no-error pass rate. In H&N plans the pass rate of all detector systems decreased with increasing magnitude of errors in the gantry angle dependent dose error scenario (Fig. 5).
3.3. Errors detected by the dosimetric systems and their clinical significance A summary of the minimum magnitude of errors detected by the three studied dosimetric systems in the considered error scenarios in prostate and H&N plans with 2%G/2 mm tolerance criteria are shown in Table 3. The Wilcoxon signed rank test statistic and p value for the minimum detected error value for each of the detector systems in the considered scenarios are shown in the same table. A p value equal or less than the value observed for the minimum detected error magnitudes shown in Table 3 were observed for the error values greater than the minimum detected errors. The dose difference in PTV D95 that corresponds to the minimum error detected by each system in each considered error scenario is also shown in Table 3. The Delta4 and Octavius 4D systems detected a minimum of 2 mm gantry angle independent and dependent MLC errors in both prostate and H&N plans. The Multicube system detected a mini-
The slope of the c pass rate (S) of each of the studied dosimetric systems in the considered error scenarios is shown in Fig. 6. In the MLC error scenario the Delta4 system showed the highest sensitivity, high magnitude of slope, in both prostate and H&N plans. In the gantry angle dependent dose error scenario the Octavius4D system showed high sensitivity. In general the Multicube system showed least sensitivity among the studied detectors systems except for the gantry angle independent MLC error scenario in H&N plans. 4. Discussion In this work we studied the performance of three commercial dosimetric systems, Delta4, Multicube and Octavius4D, in identifying delivery errors that may occur during the delivery of VMAT. Three types of errors in MLC position and relative weights of the CPs were simulated both in treatment plans and deliveries to study the clinical significance of varying magnitude of errors and the minimum magnitude of errors which could be detected for the considered systems. The dosimetric systems studied in this work have a different geometrical configuration of detector elements in the phantom; as a consequence of this the dose matrix resulting from the measurement by each system is unique in geometry and varies in the level of dose information available for the validation of treatment delivery. To account for the variability of plan complexity among plans a cohort of seven prostate and seven H&N VMAT plans that represented typical clinical plans were considered in the study. One of the major limitations of the modern electronic dosimetric systems that provide 2D or >2D dose information of the delivered dose, compared to gel dosimeters, is their lack of fine spatial resolution. The dosimetric systems considered in this study have a detector pitch ranging from 5 mm to 10 mm with a detection inactive region ranging from 3.1 mm to 10 mm (Table 2). Despite the coarse resolution of active elements the studied detector systems were shown to detect a minimum gantry angle independent and dependent MLC error of 2 mm with 2%G/2 mm tolerance criteria in both
Table 3 Summary of minimum magnitude of errors detected by the studied dosimetric systems with 2%G/2 mm tolerance criteria using Wilcoxon signed-rank test and their respective clinical significance in prostate and H&N VMAT plans. Treatment site
Error scenario
Detector Systems Delta4 Error magnitude
Prostate
H&N
Gantry angle independent MLC error Gantry angle dependent MLC error Gantry angle dependent dose error
±2 mm (W = 0, p = 0.02201) 2 mm (W = 0, p = 0.02201) 3% (W = 2, p = 0.04688)
Gantry angle independent MLC error Gantry angle dependent MLC error Gantry angle dependent dose error
±2 mm (W = 0, p = 0.02225) 2 mm (W = 0, p = 0.01563) 2% (W = 0, p = 0.01563)
Multicube Dose difference 1.4(1.0) 1.2(1.7) 0.0(0.7) 1.0(1.2) 0.0(0.7) 0.1(0.4)
Error magnitude ±1 mm (W = 0, p = 0.02201) 1 mm (W = 0 p = 0.02225) 3% (W = 1 p = 0.03125) ±1 mm (W = 0, p = 0.01788) 2 mm (W = 0, p = 0.01563) 2% (W = 0, p = 0.01563)
Octavius4D Dose difference 0.3(1.2) 0.2(0.8) 0.0(0.7) 0.3(1.1) 0.0(0.7) 0.1(0.4)
Error magnitude ±2 mm (W = 0, p = 0.02201) 2 mm (W = 0, p = 0.01563) 3% (W = 0, p = 0.01563) ±2 mm (W = 0, p = 0.02201) 2 mm (W = 0, p = 0.01563) 2% (W = 0, p = 0.01563)
Dose difference 1.4(1.0) 1.2(1.7) 0.0(0.7) 1.0(1.2) 0.0(0.7) 0.1(0.4)
W = Wilcoxon test statistic.
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
S. Arumugam et al. / Physica Medica xxx (2016) xxx–xxx
Slope of gamma pass rate (%-pt per mm for MLC errors and %-pt % dose for dose error)
6
-12 Delta4 Multicube Octavius4D
-10 -8 -6 -4 -2 0
Gantry angle Gantry angle Gantry angle Gantry angle Gantry angle Gantry angle dependent dependent independent dependent dependent independent dose error MLC error MLC error dose error MLC error MLC error Prostate
H&N
Treatment site and type of error Fig. 6. The slope of gamma pass rate line of Delta4, Multicube and Octavius4D dosimetric systems in prostate and H&N VMAT plans for the studied error scenarios.
prostate and H&N plans. This 2 mm error corresponds to a maximum of 1.4(1.0)% and 1.0(1.2)% decrease in PTV D95 for Prostate and H&N plans respectively (Table 3). The relatively smaller detector inactive region (3.1 mm) of the Multicube detector could be the reason for its ability to detect a smaller gantry angle independent MLC error (1 mm). The Wilcoxon signed-rank test used in this study identifies the statistical difference in c pass rate between no error and error introduced plans. A statistically significant difference in pass rate between no error and error plans was observed even at lower error magnitudes as the pass rate of error introduced plans are always lower than their respective no error plans. While in the controlled condition of the study it is relatively easy to assess the minimum error detected by the system, by comparing it with the respective no error plans, in a clinical QA scenario this is challenging as a relatively wide range of c pass rates are observed for a given treatment site especially with tighter tolerance criteria. This is supported by the pass rates observed for the Multicube system in the gantry angle dependent dose error scenario for prostate plans (Fig. 5). Though a statistically significant difference in pass rate between no error and dose error plans was observed for prostate plans, due to the lower sensitivity of the detector for this particular type of error in prostate plans a mean pass rate of 88.7(3.1)% was observed with 10% dose error while the mean pass rate for the no error plans is 92.1(1.3)% (Fig. 5). Fredth et al. [11] studied the ability of Delta4, Octavius, COMPASS and Epiqa systems in detecting simulated machine output and MLC calibration errors in VMAT delivery. Their study showed that the Epiqa system detected all the simulated errors and the Delta 4 system detected 15 out of 20 simulated errors, whilst the Compass and Octavius systems detected 8 out of 20 errors. Also they concluded that the ability of a particular detector system in detecting error is specific to the treatment plan. In this study we systematically studied the ability of the detector systems in identifying minimum magnitude of error in a cohort of plans with high and low complexity levels. Similarly Vieillevigne et al. [20] studied the sensitivity of dose matrix by ArcCHECK, 2d array729 and EPID devices for systematic delivery errors in collimator, gantry and couch angles and MLC position. Vieillevigne et al. concluded that all the studied systems had the same level of threshold in detecting the errors. In the presented work we investigated the minimum error detected by three dosimetric systems which have detector elements with different physical characteristics and geometrical arrangement that influence the measured dose and spatial resolu-
tion and the ability to measure data ranging from single planar dose matrix to 3D dose matrix. As demonstrated by our results, statistically all three dosimetric systems showed a similar magnitude of minimum error detection capability. However the relative sensitivity of the detector system to simulated errors is shown to be relatively higher for less complex plans, in the multi-planar detector systems such as Delta4 and Octavius 4D, in comparison to single planar detector (Multicube).
5. Conclusion The sensitivity of Delta4, Multicube and Octavius 4D systems in detecting the VMAT delivery errors was studied by introducing three types of gantry angle independent and dependent MLC and dose errors. All three dosimetric systems detected the angle independent MLC errors that would result in a clinically significant impact on D95 of PTV in prostate and H&N plans; a minimum gantry angle independent and dependent MLC error of 2 mm was detected by all the systems. In the gantry angle dependent dose error scenario all three systems detected the error of 3% and 2% in prostate and H&N plans respectively.
Acknowledgement This project was funded through a Cancer Council NSW Project Grant (RG14-11).
References [1] Otto K. Volumetric modulated arc therapy: IMRT in a single gantry arc. Med Phys 2008;35(1):310–7. [2] Ezzell GA, Galvin JM, Low D, Palta JR, Rosen I, Sharpe MB, Xia P, Xiao Y, Xing L, Cedric XY. Guidance document on delivery, treatment planning, and clinical implementation of IMRT: report of the IMRT Subcommittee of the AAPM Radiation Therapy Committee. Med Phys 2003;30(8):2089–115. [3] Baldock C. Historical overview of the development of gel dosimetry: another personal perspective. J Phys: Conf Ser 2009;164(012002):1–20. [4] Baldock C, De Deene Y, Doran S, Ibbott G, Jirasek A, Lepage M, et al. Polymer gel dosimetry. Phys Med Biol 2010;55(5):R1. [5] Feygelman V, Zhang G, Stevens C, Nelms BE. Evaluation of a new VMAT QA device, or the ‘‘X” and ‘‘O” array geometries. J Appl Clin Med Phys 2011;12(2). [6] Boggula R, Birkner M, Lohr F, Steil V, Wenz F, Wertz H. Evaluation of a 2D detector array for patient-specific VMAT QA with different setups. Phys Med Biol 2011;56(22):7163. [7] Wolfsberger L, Wagar M, Nitsch P, Bhagwat M, Zygmanski P. Angular dose dependence of Matrixx TM and its calibration. J Appl Clin Med Phys 2010;11 (1).
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
S. Arumugam et al. / Physica Medica xxx (2016) xxx–xxx [8] Van Esch A, Basta K, Evrard M, Ghislain M, Sergent F, Huyskens DP. The Octavius1500 2D ion chamber array and its associated phantoms: dosimetric characterization of a new prototype. Med Phys 2014;41(9):091708. [9] Bedford JL, Lee YK, Wai P, South CP, Warrington AP. Evaluation of the Delta4 phantom for IMRT and VMAT verification. Phys Med Biol 2009;54 (9):N167. [10] Negri A, Scaggion A, Rossato M, Canonico D, Zandonà R, Paiusco M. Evaluation of a phantom related gamma index threshold for VMAT QA. Phys Med 2014;30:e70. [11] Fredh A, Scherman JB, Fog LS, Af Rosenschöld PM. Patient QA systems for rotational radiation therapy: a comparative experimental study with intentional errors. Med Phys 2013;40(3):031716. [12] Arumugam S, Xing A, Young T, Holloway L. Sensitivity of a helical diode array dosimeter to Volumetric Modulated Arc Therapy delivery errors. Phys Med 2015;31(8):1043–54. [13] Feygelman V, Zhang GG, Stevens CW. Initial dosimetric evaluation of SmartArc-a novel VMAT treatment planning module implemented in a multi-vendor delivery chain. J Appl Clin Med Phys 2010;11(1).
7
[14] Masi L, Doro R, Favuzza V, Cipressi S, Livi L. Impact of plan parameters on the dosimetric accuracy of volumetric modulated arc therapy. Med Phys 2013;40 (7):071718. [15] Arumugam S, Xing A, Goozee G, Holloway L. Detecting VMAT delivery errors: A study on the sensitivity of the ArcCHECK-3D electronic dosimeter. J Phys: Conf Ser 2013;444(1):012019. [16] Stathakis S, Myers P, Esquivel C, Mavroidis P, Papanikolaou N. Characterization of a novel 2D array dosimeter for patient-specific quality assurance with volumetric arc therapy. Med Phys 2013;40(7):071731. [17] Low DA, Dempsey JF. Evaluation of the gamma dose distribution comparison method. Med Phys 2003;30(9):2455–64. [18] Wilcoxon F. Individual comparisons by ranking methods. Biometrics Bull 1945;1(6):80–3. [19] Ihaka R, Gentleman R. R: a language for data analysis and graphics. J Comput Graphics Stat 1996;5(3):299–314. [20] Vieillevigne L, Molinier J, Brun T, Ferrand R. Gamma index comparison of three VMAT QA systems and evaluation of their sensitivity to delivery errors. Phys Med 2015;31(7):720–5.
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
Contents lists available at ScienceDirect
Physica Medica journal homepage: http://www.physicamedica.com
Original paper
Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors Sankar Arumugam a,b,⇑, Aitang Xing a,b, Tony Young a,c, David Thwaites a,c, Lois Holloway a,b,c,d a
Liverpool and Macarthur Cancer Therapy Centres and Ingham Institute, Liverpool, New South Wales, Australia University of New South Wales, Sydney, New South Wales, Australia c Institute of Medical Physics, School of Physics, University of Sydney, Sydney, New South Wales, Australia d Centre for Medical Radiation Physics, University of Wollongong, Wollongong, New South Wales, Australia b
a r t i c l e
i n f o
Article history: Received 31 May 2016 Received in Revised form 23 August 2016 Accepted 22 September 2016 Available online xxxx Keywords: VMAT Dosimetric systems Delivery error
a b s t r a c t Aim: To study the sensitivity of three commercial dosimetric systems, Delta4, Multicube and Octavius4D, in detecting Volumetric Modulated Arc Therapy (VMAT) delivery errors. Methods: Fourteen prostate and head and neck (H&N) VMAT plans were considered for this study. Three types of errors were introduced into the original plans: gantry angle independent and dependent MLC errors, and gantry angle dependent dose errors. The dose matrix measured by each detector system for the no-error and error introduced delivery were compared with the reference Treatment Planning System (TPS) calculated dose matrix for no-error plans using gamma (c) analysis with 2%/2 mm tolerance criteria. The ability of the detector system in identifying the minimum error in each scenario was assessed by analysing the gamma pass rates of no error delivery and error delivery using a Wilcoxon signed-rank test. The relative sensitivity of the system was assessed by determining the slope of the gamma pass line for studied error magnitude in each error scenario. Results: In the gantry angle independent and dependent MLC error scenario the Delta4, Multicube and Octavius4D systems detected a minimum 2 mm error. In the gantry angle dependent dose error scenario all studied systems detected a minimum 3% and 2% error in prostate and H&N plans respectively. In the studied detector systems Multicube showed relatively less sensitivity to the errors in the majority of error scenarios. Conclusion: The studied systems identified the same magnitude of minimum errors in all considered error scenarios. Ó 2016 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.
1. Introduction Volumetric Modulated Arc Therapy (VMAT) generates highly conformal radiotherapy plans with reduced treatment delivery time [1]. These advantages come with added complexity at planning and delivery stages of the treatment. The plan specific dosimetric quality assurance (QA) check is an integral part of the treatment delivery chain in these techniques to ensure the accuracy of dose delivery [2]. The rotational delivery aspect of these treatment techniques demands dosimetric systems that are insensitive to beam incident angle and provide delivered dose information in multi dimensions with high spatial resolution.
⇑ Corresponding author at: Department of Radiation Oncology, Cancer Therapy Centre, Liverpool Hospital, Sydney, NSW 2170, Australia. E-mail address: [email protected] (S. Arumugam).
An ideal dosimetric system for the verification of advanced treatment techniques would provide high resolution 3D spatial information of the delivered dose. Gel dosimeters are well-suited for this purpose [3,4], however their widespread use in routine clinical practice is limited due to various factors such as the requirement of a high level of expertise, the requirement of expensive readout systems and high sensitivity of gel dosimeters to gel preparation, composition and ambient conditions. Currently available electronic dosimetric systems provide a reconstructed measured dose matrix for rotational treatment delivery by accounting for appropriate corrections to the beam incident angle on the detector elements available [5–7]. The detector element arrangements in modern electronic dosimeters vary widely; the systems have configurations with stationary and rotating 2D planar arrays, biplanar 2D arrays and helical grid arrays in an annular plane of a cylindrical phantom [5,6,8]. The measured dose matrix dimension varies from a two
http://dx.doi.org/10.1016/j.ejmp.2016.09.016 1120-1797/Ó 2016 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
2
S. Arumugam et al. / Physica Medica xxx (2016) xxx–xxx
dimensional matrix that measures single planar dose to multiplanar and complete three dimensional matrices depending on both the geometry of the detector arrangement and data processing used. Previous studies have investigated intrinsic characteristics, utilisation and limitations of modern detector systems in the dosimetric QA of advanced radiotherapy techniques [5,6,8–10]. Fredh, et al. [11] investigated four commercial dosimetric systems and their response to MLC bank position and dosimetric and collimator errors introduced to VMAT plans. Fredh et al. concluded that the sensitivity of each particular detector system’s ability to detect errors is specific to a given radiotherapy plan. Their study didn’t assess the relative sensitivity of the detector systems in identifying the errors. The purpose of this work is to study the ability of three commercial detector systems, Delta4, Multicube and Octavius 4D, in detecting delivery errors that are characteristic to VMAT. The clinical significance of minimum error detected by each dosimetric system was assessed by comparing the previously published dosimetric impact of the errors calculated in the Treatment Planning System (TPS) for the same data set by our research group [12].
2.3. Studied errors Three types of errors were intentionally introduced to the original plans considered in this study. In-house developed software was used to introduce the errors [12,15]. 1. Gantry angle independent MLC shift to simulate the systematic offset in the MLC position (ranging from ±1 mm to ±5 mm). 2. Gantry angle dependent MLC error (amplitude ranging from 1 mm to 5 mm), to simulate the sag in MLC position as a function of gantry angle [12]. 3. Gantry angle dependent dose error (amplitude ranging from 1% to 10%) to simulate the output fluctuation as a function of gantry angle [12]. The total monitor units (MU) of the VMAT arcs were maintained with the specified gantry angle dependent errors introduced to the relative weights of the control points (CPs). The error introduced plans were recalculated in the TPS and the treatment delivery files of these plans were exported to the LA for delivery.
2. Materials and methods
2.4. Clinical significance of introduced errors
2.1. Planning and delivery
The impact of introduced errors on the dose to the planning target volume (PTV) and organs at risk (OARs) was published in our previous work [12]. For the convenience of correlating the error detected by the dosimetric systems considered in this study the results are compared here. For prostate plans the impact of the studied errors on the dose delivered to 95% of PTV (D95), volume receiving 60 Gy (V60) for rectum and dose delivered to 50% (D50) of bladder was assessed. Similarly in H&N plans D95 to high dose PTV (HD-PTV), D95 to low dose PTV (LD-PTV) and the maximum dose delivered to a 2 cc volume (D2cc) of the spinal cord were considered. The percentage change in studied dose metrics with respect to the no-error plans was quantified to study the impact of errors.
The VMAT plans were generated using the SmartArc optimisation tool available in the Pinnacle v9.8 (Philips Healthcare, Fitchburg, WI, USA) Treatment Planning System (TPS). A 6 MV photon beam model for an Elekta-Synergy (Elekta Ltd, Crawley, UK) linear accelerator (LA) was used for planning. The details of the SmartArc optimisation algorithm used in Pinnacle can be found elsewhere [13]. Dose calculations for all plans in the study were performed using the Adaptive Convolution dose calculation algorithm available in the Pinnacle TPS with 2.5 2.5 2.5 mm3 voxel size. The structures representing the Elekta Evo table top with appropriate densities were included in the dose calculation in all patient and verification plans. The Synergy accelerator used in this study was equipped with an MLCi head and desktop pro, v 7.0, LA control software. Mosaiq, v 2.30.0D1, (Impac Medical Systems, Inc. California, USA) record and verification (R&V) system was used for the transfer of VMAT plans from TPS to LA.
2.2. VMAT plans Fourteen clinical VMAT plans (Seven each of prostate and head and neck (H&N)) were considered for the study. Some of the key characteristics of the plans are shown in Table 1. To assess the complexity of the plans considered in the study a multiplicative combination of leaf travel and modulation complexity score [14] (LT-MCS) was calculated using in-house software [14]. The mean (r) LT-MCS value for simple arc, prostate and H&N plans are shown in Table 1. The lower LT-MCS value represents higher complexity in the plan. As seen in Table 1 H&N plans were highly complex for the studied plan cohorts.
Table 1 Some of the important characteristics of the VMAT plans considered. Parameter
Arc length Coll angle Mean (r)LT-MSCv
Parameter value Prostate plans
H&N plans
240°–352° 10° 0.4(0.035)
352° 10° 0.07(0.05)
2.5. Studied dosimetric systems The following three commercial dosimetric systems were assessed for their ability to detect the errors considered in each error scenario. Table 2 shows some of the key characteristics of these systems. 2.5.1. Delta4 The Delta4 (ScandiDos AB, Uppsala, Sweden) is a cylindrical acrylic phantom containing 1069 diode detectors arranged in two orthogonal planes. The measurement planes pass through the centre of the phantom and are 45° oblique to the sagittal and coronal planes. Table 2 summarises some of the key characteristics of the Delta4 system and a more detailed description can be found elsewhere [5]. The measured signal is corrected for background, beam incident angle and ambient temperature. The manufacturer recommended dose calibration procedure was followed to correct for the detector array relative response and dose calibration. The verification treatment plans were recalculated on the pseudo CT data set representing the Delta4 geometry. All the dose calculations were performed with a density of 1.19 g/cm3 forced to Delta4 geometry in the TPS. ScandiDos Delta4 software (ScandiDos AB, Uppsala, Sweden) was used for the measurement and analysis of agreement to TPS calculated dose. 2.5.2. MatrixxEvolution with MultiCube MatrixxEvolution (IBA Dosimetry GmbH, Germany) is an ionchamber 2D array consisting of 1020 cylindrical shaped
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
3
S. Arumugam et al. / Physica Medica xxx (2016) xxx–xxx Table 2 The key physical characteristics and measured dose matrix of the dosimetric systems considered in the study. Detector systems characteristics
Number of detector elements and their type and arrangement Detector size Detector pitch Detector inactive region Final dose matrix
Detector system Delta4
Multicube
Octavius4D
1069 diode detectors Orthogonal planes 0.78 mm2 5 mm at central 6 6 cm2 and 10 mm outer 20 20 cm2 5 mm at central 6 6 cm2 and 10 mm outer 20 20 cm2 Orthogonal planes (with the option of 3D dose matrix)
1020 ion chambers Coronal plane 4.5 mm/ and 5 mm height 7.62 mm 3.1 mm Coronal plane
729 ion chambers 2D array in rotating phantom 5 mm 5 mm 5 mm 10 mm 5 mm 3D dose matrix
ion-chambers arranged in a 2D matrix with a 32 32 grid (Table 2). The MultiCube (IBA Dosimetry GmbH, Germany) is a cubical phantom of 30 cm 30 cm 25 cm dimension and can house the MatrixxEvolution which enables a planar dose measurement in the coronal plane. A detailed description of the MatrixxEvolution and its application in the quality assurance of rotational treatment with MultiCube can be found elsewhere [7]. The measured signal is corrected for background, angular response of the ion chambers and ambient temperature and pressure changes from the calibration measurement. The manufacturer provided calibration file was used to correct the response differences of the ion chambers. The absolute calibration of the device was performed as per the manufacturer’s recommendation. The verification treatment plans were recalculated on the CT data set of the MultiCube phantom with the MatrixxEvolution. The planar dose corresponding to the measurement plane of the MatrixxEvolution was exported from the TPS for comparison with measurement. The Omnipro I’mRT (v 6.1) software (IBA Dosimetry GmbH, Germany) was used to perform the measurements and comparison of measured and TPS dose matrices. 2.5.3. Octavius 4D 729 Octavius 4D 729 (PTW Freiburg GmbH, Germany) consists of a PTW 729 ion chamber 2D array housed in an Octavius 4D phantom. The PTW 729 detector array consists of 729 cubical ion chambers arranged in 2D matrix with a 27 27 grid. The Octavius4D phantom is a cylindrical phantom of 32 cm diameter and 34.3 cm length. The phantom rotates in synchrony with the gantry of the LA so that the measurement plane of the 2D array is always perpendicular to the beam axis. The gantry angle sorted 2D dose matrices are used for the reconstruction of 3D matrices with Percentage Depth Dose (PDD) data of the treatment beam measured at 85 cm SSD. The cumulative 3D dose matrix of the measured plan is calculated by adding all the reconstructed dose matrices. More details on the description and function of the Octavius 4D phantom and the PTW 729 array detector can be found elsewhere [16]. The signal measured by the PTW 729 array detector is corrected for background and ambient temperature and pressure differences from calibration measurements. The manufacturer provided calibration file was used to correct the response differences of the ion chambers in the 2D array. The absolute calibration of the device was performed as per the manufacturer’s recommendation. The verification treatment plans were recalculated on the pseudo CT data set representing a cylinder of dimensions equivalent to the Octavius4D phantom. The dose calculations were performed by forcing the density of 1.05 g/cm3 for the phantom. The Verisoft (v5.1) software (PTW Freiburg GmbH, Germany) was used to perform the measurements and 3D dose matrices were reconstructed with the voxel size of 2.5 mm 2.5 mm 2.5 mm. The comparison of TPS and reconstructed dose matrices was performed in Verisoft software. 2.6. Dose matrix comparison The measured and TPS calculated dose matrices were compared using c analysis [17] with 2% global dose tolerance and 2 mm
distance-to-agreement (DTA) tolerance criteria (2%G/2 mm). The maximum dose in TPS dose matrices of respective detector geometry was considered. The verification software included in the relevant dosimetric systems was used for the analysis of TPS and respective measured dose matrices. The dose matrix elements that receive less than 10% of the mean dose to PTV in respective detector geometry were not considered in the analysis. 2.7. Data analysis The minimum error detected by each of the dosimetric systems in prostate and H&N plans was determined by performing a Wilcoxon signed-rank test [18] between paired data of c pass rate of no error and error plans. The R statistical computing and graphics software was used for the calculation of the Wilcoxon test statistic (W) and p value [19]. In the Wilcoxon signed-rank analysis of data the minimum magnitude of a particular error was considered detected by a dosimetric system if the test statistic suggested a statistical difference (p < 0.05) between the paired no error and error pass rates. 2.8. Relative sensitivity of the systems to delivery errors The sensitivity of each dosimetric system to delivery errors was studied by determining the slope of the line (S) describing the gamma pass rate for no error and the studied range of errors in each error scenario. For the purpose of calculating slope, the gamma pass rate of each system for the studied error range was fitted using a linear least square fit. 3. Results 3.1. Impact of introduced errors on dose-volume metrics Figs. 1 and 2 shows the mean (r)% change in dose-volume metrics for the studied error scenarios in the prostate and H&N plans respectively. In both prostate and H&N plans the MLC errors resulted in the decrease in PTV D95 and the magnitude of decrease was proportional to the magnitude of error. In the gantry angle dependent dose error scenario the PTV D95 in prostate plans was not affected by the error whereas a small change was observed in H&N plans. 3.2. Change in gamma pass rate with errors The mean (r) c pass rate for no error and error introduced plans by each of the detector systems for prostate and H&N plans are shown in Figs. 3–5 for gantry angle independent MLC error, gantry angle dependent MLC error and dose error scenarios respectively. In both prostate and H&N plans the pass rate of all the detectors decreased, compared to no error plans, as the magnitude of MLC error increased in both gantry angle dependent and independent MLC error scenarios. In the gantry angle dependent dose error sce-
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
4
S. Arumugam et al. / Physica Medica xxx (2016) xxx–xxx
PTV D95 Bladder D50 Rectum V60
10.0 5.0 0.0 -5.0
-10.0 -15.0
No error 1 2 3 4 5 7 10
No error 1 2 3 4 5
No error ±1 ±2 ±3 ±4 ±5
No error 1 2 3 4 5 7 10
-25.0
No error 1 2 3 4 5
-20.0 No error ±1 ±2 ±3 ±4 ±5
% difference in DVH metric
15.0
Gantry angle Gantry angle Gantry angle dependent Gantry angle Gantry angle Gantry angle independent MLC dependent MLC dose error (%) independent MLC dependent MLC dependent dose error error (mm) error (mm) error (mm) error (mm) (%) PTV
OARs
Fig. 1. The mean (r) percentage change in PTV D95, Rectum V60 and Bladder D50 for the studied error scenarios in Prostate VMAT plans.
LD-PTV D95 HD-PTV D95
10.0
Spinal cord D2cc 5.0 0.0
No error 1 2 3 4 5 7 10
No error 1 2 3 4 5
No error ±1 ±2 ±3 ±4 ±5
No error 1 2 3 4 5 7 10
-10.0
No error 1 2 3 4 5
-5.0
No error ±1 ±2 ±3 ±4 ±5
% difference in DVH metric
15.0
Gantry angle Gantry angle Gantry angle dependent Gantry angle Gantry angle Gantry angle independent MLC dependent MLC dose error (%) independent MLC dependent MLC dependent dose error error (mm) error (mm) error (mm) error (mm) (%) PTV
OAR
Fig. 2. The mean (r) percentage change in D95 of High dose PTV (HD-PTV), Low dose PTV (LD-PTV), and Spinal cord D2cc for the studied error scenarios in Head and Neck VMAT plans.
100
90 80 70 60
Delta4 Multicube
90.0
Octavius4D pass rate (%)
pass rate (%)
100.0
Delta4 Multicube Octavius4D
80.0 70.0 60.0
50
50.0 40
No ±1 error
±2
±3 Prostate
±4
±5
No ±1 error
±2
±3
±4
±5
H&N
Treatment site and MLC error (mm) Fig. 3. The mean (r) gamma pass rates for no error and error introduced delivery in the gantry angle independent MLC error scenario for prostate and H&N plans.
40.0 No 1 error
2
3 Prostate
4
5
No 1 error
2
3
4
5
H&N
Treatment site and MLC error (mm) Fig. 4. The mean (r) gamma pass rates for no error and error introduced delivery in the gantry angle dependent MLC error scenario for prostate and H&N plans.
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
5
S. Arumugam et al. / Physica Medica xxx (2016) xxx–xxx
Delta4
100.0
Multicube
pass rate (%)
90.0
Octavius4D
80.0
mum 1 mm error in both MLC error scenarios in prostate plans and detected 1 mm and 2 mm errors in gantry angle independent and dependent MLC error scenarios in H&N plans. All three systems detected 3% gantry angle dependent dose errors in prostate plans and 2% error in H&N plans.
70.0 60.0
3.4. Relative sensitivity of the dosimetric systems in detecting errors
50.0 40.0
Prostate
10
7
No error 1 2 3 4 5
10
7
No error 1 2 3 4 5
30.0
H&N
Treatment site and dose error (%) Fig. 5. The mean (r) gamma pass rates for no error and error introduced delivery in the gantry angle dependent dose error scenario for prostate and H&N plans.
nario the c pass rate of the Multicube system showed minimum change (only 4% reduction from no error mean pass rate for 10% dose error) compared to no-error pass rate. In H&N plans the pass rate of all detector systems decreased with increasing magnitude of errors in the gantry angle dependent dose error scenario (Fig. 5).
3.3. Errors detected by the dosimetric systems and their clinical significance A summary of the minimum magnitude of errors detected by the three studied dosimetric systems in the considered error scenarios in prostate and H&N plans with 2%G/2 mm tolerance criteria are shown in Table 3. The Wilcoxon signed rank test statistic and p value for the minimum detected error value for each of the detector systems in the considered scenarios are shown in the same table. A p value equal or less than the value observed for the minimum detected error magnitudes shown in Table 3 were observed for the error values greater than the minimum detected errors. The dose difference in PTV D95 that corresponds to the minimum error detected by each system in each considered error scenario is also shown in Table 3. The Delta4 and Octavius 4D systems detected a minimum of 2 mm gantry angle independent and dependent MLC errors in both prostate and H&N plans. The Multicube system detected a mini-
The slope of the c pass rate (S) of each of the studied dosimetric systems in the considered error scenarios is shown in Fig. 6. In the MLC error scenario the Delta4 system showed the highest sensitivity, high magnitude of slope, in both prostate and H&N plans. In the gantry angle dependent dose error scenario the Octavius4D system showed high sensitivity. In general the Multicube system showed least sensitivity among the studied detectors systems except for the gantry angle independent MLC error scenario in H&N plans. 4. Discussion In this work we studied the performance of three commercial dosimetric systems, Delta4, Multicube and Octavius4D, in identifying delivery errors that may occur during the delivery of VMAT. Three types of errors in MLC position and relative weights of the CPs were simulated both in treatment plans and deliveries to study the clinical significance of varying magnitude of errors and the minimum magnitude of errors which could be detected for the considered systems. The dosimetric systems studied in this work have a different geometrical configuration of detector elements in the phantom; as a consequence of this the dose matrix resulting from the measurement by each system is unique in geometry and varies in the level of dose information available for the validation of treatment delivery. To account for the variability of plan complexity among plans a cohort of seven prostate and seven H&N VMAT plans that represented typical clinical plans were considered in the study. One of the major limitations of the modern electronic dosimetric systems that provide 2D or >2D dose information of the delivered dose, compared to gel dosimeters, is their lack of fine spatial resolution. The dosimetric systems considered in this study have a detector pitch ranging from 5 mm to 10 mm with a detection inactive region ranging from 3.1 mm to 10 mm (Table 2). Despite the coarse resolution of active elements the studied detector systems were shown to detect a minimum gantry angle independent and dependent MLC error of 2 mm with 2%G/2 mm tolerance criteria in both
Table 3 Summary of minimum magnitude of errors detected by the studied dosimetric systems with 2%G/2 mm tolerance criteria using Wilcoxon signed-rank test and their respective clinical significance in prostate and H&N VMAT plans. Treatment site
Error scenario
Detector Systems Delta4 Error magnitude
Prostate
H&N
Gantry angle independent MLC error Gantry angle dependent MLC error Gantry angle dependent dose error
±2 mm (W = 0, p = 0.02201) 2 mm (W = 0, p = 0.02201) 3% (W = 2, p = 0.04688)
Gantry angle independent MLC error Gantry angle dependent MLC error Gantry angle dependent dose error
±2 mm (W = 0, p = 0.02225) 2 mm (W = 0, p = 0.01563) 2% (W = 0, p = 0.01563)
Multicube Dose difference 1.4(1.0) 1.2(1.7) 0.0(0.7) 1.0(1.2) 0.0(0.7) 0.1(0.4)
Error magnitude ±1 mm (W = 0, p = 0.02201) 1 mm (W = 0 p = 0.02225) 3% (W = 1 p = 0.03125) ±1 mm (W = 0, p = 0.01788) 2 mm (W = 0, p = 0.01563) 2% (W = 0, p = 0.01563)
Octavius4D Dose difference 0.3(1.2) 0.2(0.8) 0.0(0.7) 0.3(1.1) 0.0(0.7) 0.1(0.4)
Error magnitude ±2 mm (W = 0, p = 0.02201) 2 mm (W = 0, p = 0.01563) 3% (W = 0, p = 0.01563) ±2 mm (W = 0, p = 0.02201) 2 mm (W = 0, p = 0.01563) 2% (W = 0, p = 0.01563)
Dose difference 1.4(1.0) 1.2(1.7) 0.0(0.7) 1.0(1.2) 0.0(0.7) 0.1(0.4)
W = Wilcoxon test statistic.
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
S. Arumugam et al. / Physica Medica xxx (2016) xxx–xxx
Slope of gamma pass rate (%-pt per mm for MLC errors and %-pt % dose for dose error)
6
-12 Delta4 Multicube Octavius4D
-10 -8 -6 -4 -2 0
Gantry angle Gantry angle Gantry angle Gantry angle Gantry angle Gantry angle dependent dependent independent dependent dependent independent dose error MLC error MLC error dose error MLC error MLC error Prostate
H&N
Treatment site and type of error Fig. 6. The slope of gamma pass rate line of Delta4, Multicube and Octavius4D dosimetric systems in prostate and H&N VMAT plans for the studied error scenarios.
prostate and H&N plans. This 2 mm error corresponds to a maximum of 1.4(1.0)% and 1.0(1.2)% decrease in PTV D95 for Prostate and H&N plans respectively (Table 3). The relatively smaller detector inactive region (3.1 mm) of the Multicube detector could be the reason for its ability to detect a smaller gantry angle independent MLC error (1 mm). The Wilcoxon signed-rank test used in this study identifies the statistical difference in c pass rate between no error and error introduced plans. A statistically significant difference in pass rate between no error and error plans was observed even at lower error magnitudes as the pass rate of error introduced plans are always lower than their respective no error plans. While in the controlled condition of the study it is relatively easy to assess the minimum error detected by the system, by comparing it with the respective no error plans, in a clinical QA scenario this is challenging as a relatively wide range of c pass rates are observed for a given treatment site especially with tighter tolerance criteria. This is supported by the pass rates observed for the Multicube system in the gantry angle dependent dose error scenario for prostate plans (Fig. 5). Though a statistically significant difference in pass rate between no error and dose error plans was observed for prostate plans, due to the lower sensitivity of the detector for this particular type of error in prostate plans a mean pass rate of 88.7(3.1)% was observed with 10% dose error while the mean pass rate for the no error plans is 92.1(1.3)% (Fig. 5). Fredth et al. [11] studied the ability of Delta4, Octavius, COMPASS and Epiqa systems in detecting simulated machine output and MLC calibration errors in VMAT delivery. Their study showed that the Epiqa system detected all the simulated errors and the Delta 4 system detected 15 out of 20 simulated errors, whilst the Compass and Octavius systems detected 8 out of 20 errors. Also they concluded that the ability of a particular detector system in detecting error is specific to the treatment plan. In this study we systematically studied the ability of the detector systems in identifying minimum magnitude of error in a cohort of plans with high and low complexity levels. Similarly Vieillevigne et al. [20] studied the sensitivity of dose matrix by ArcCHECK, 2d array729 and EPID devices for systematic delivery errors in collimator, gantry and couch angles and MLC position. Vieillevigne et al. concluded that all the studied systems had the same level of threshold in detecting the errors. In the presented work we investigated the minimum error detected by three dosimetric systems which have detector elements with different physical characteristics and geometrical arrangement that influence the measured dose and spatial resolu-
tion and the ability to measure data ranging from single planar dose matrix to 3D dose matrix. As demonstrated by our results, statistically all three dosimetric systems showed a similar magnitude of minimum error detection capability. However the relative sensitivity of the detector system to simulated errors is shown to be relatively higher for less complex plans, in the multi-planar detector systems such as Delta4 and Octavius 4D, in comparison to single planar detector (Multicube).
5. Conclusion The sensitivity of Delta4, Multicube and Octavius 4D systems in detecting the VMAT delivery errors was studied by introducing three types of gantry angle independent and dependent MLC and dose errors. All three dosimetric systems detected the angle independent MLC errors that would result in a clinically significant impact on D95 of PTV in prostate and H&N plans; a minimum gantry angle independent and dependent MLC error of 2 mm was detected by all the systems. In the gantry angle dependent dose error scenario all three systems detected the error of 3% and 2% in prostate and H&N plans respectively.
Acknowledgement This project was funded through a Cancer Council NSW Project Grant (RG14-11).
References [1] Otto K. Volumetric modulated arc therapy: IMRT in a single gantry arc. Med Phys 2008;35(1):310–7. [2] Ezzell GA, Galvin JM, Low D, Palta JR, Rosen I, Sharpe MB, Xia P, Xiao Y, Xing L, Cedric XY. Guidance document on delivery, treatment planning, and clinical implementation of IMRT: report of the IMRT Subcommittee of the AAPM Radiation Therapy Committee. Med Phys 2003;30(8):2089–115. [3] Baldock C. Historical overview of the development of gel dosimetry: another personal perspective. J Phys: Conf Ser 2009;164(012002):1–20. [4] Baldock C, De Deene Y, Doran S, Ibbott G, Jirasek A, Lepage M, et al. Polymer gel dosimetry. Phys Med Biol 2010;55(5):R1. [5] Feygelman V, Zhang G, Stevens C, Nelms BE. Evaluation of a new VMAT QA device, or the ‘‘X” and ‘‘O” array geometries. J Appl Clin Med Phys 2011;12(2). [6] Boggula R, Birkner M, Lohr F, Steil V, Wenz F, Wertz H. Evaluation of a 2D detector array for patient-specific VMAT QA with different setups. Phys Med Biol 2011;56(22):7163. [7] Wolfsberger L, Wagar M, Nitsch P, Bhagwat M, Zygmanski P. Angular dose dependence of Matrixx TM and its calibration. J Appl Clin Med Phys 2010;11 (1).
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016
S. Arumugam et al. / Physica Medica xxx (2016) xxx–xxx [8] Van Esch A, Basta K, Evrard M, Ghislain M, Sergent F, Huyskens DP. The Octavius1500 2D ion chamber array and its associated phantoms: dosimetric characterization of a new prototype. Med Phys 2014;41(9):091708. [9] Bedford JL, Lee YK, Wai P, South CP, Warrington AP. Evaluation of the Delta4 phantom for IMRT and VMAT verification. Phys Med Biol 2009;54 (9):N167. [10] Negri A, Scaggion A, Rossato M, Canonico D, Zandonà R, Paiusco M. Evaluation of a phantom related gamma index threshold for VMAT QA. Phys Med 2014;30:e70. [11] Fredh A, Scherman JB, Fog LS, Af Rosenschöld PM. Patient QA systems for rotational radiation therapy: a comparative experimental study with intentional errors. Med Phys 2013;40(3):031716. [12] Arumugam S, Xing A, Young T, Holloway L. Sensitivity of a helical diode array dosimeter to Volumetric Modulated Arc Therapy delivery errors. Phys Med 2015;31(8):1043–54. [13] Feygelman V, Zhang GG, Stevens CW. Initial dosimetric evaluation of SmartArc-a novel VMAT treatment planning module implemented in a multi-vendor delivery chain. J Appl Clin Med Phys 2010;11(1).
7
[14] Masi L, Doro R, Favuzza V, Cipressi S, Livi L. Impact of plan parameters on the dosimetric accuracy of volumetric modulated arc therapy. Med Phys 2013;40 (7):071718. [15] Arumugam S, Xing A, Goozee G, Holloway L. Detecting VMAT delivery errors: A study on the sensitivity of the ArcCHECK-3D electronic dosimeter. J Phys: Conf Ser 2013;444(1):012019. [16] Stathakis S, Myers P, Esquivel C, Mavroidis P, Papanikolaou N. Characterization of a novel 2D array dosimeter for patient-specific quality assurance with volumetric arc therapy. Med Phys 2013;40(7):071731. [17] Low DA, Dempsey JF. Evaluation of the gamma dose distribution comparison method. Med Phys 2003;30(9):2455–64. [18] Wilcoxon F. Individual comparisons by ranking methods. Biometrics Bull 1945;1(6):80–3. [19] Ihaka R, Gentleman R. R: a language for data analysis and graphics. J Comput Graphics Stat 1996;5(3):299–314. [20] Vieillevigne L, Molinier J, Brun T, Ferrand R. Gamma index comparison of three VMAT QA systems and evaluation of their sensitivity to delivery errors. Phys Med 2015;31(7):720–5.
Please cite this article in press as: Arumugam S et al. Comparison of three commercial dosimetric systems in detecting clinically significant VMAT delivery errors. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.09.016