Practical application of Octavius®‐4D: Characteristics and criticalities for IMRT and VMAT verification

Abstract Octavius®‐4D is a very effective device in radiotherapy treatment quality assurance (QA), due to its simple set‐up and analysis package. However, even if it is widely used, its main characteristics and criticalities were only partially investigated. Taking start from its commissioning, the aim of this work was to study the main dependencies of the device response. The outcome dependence was studied comparing results by different delivery techniques [Intensity Modulated Radiation Therapy, IMRT (n = 29) and RapidArc, RA (n = 15)], anatomical regions [15 head/neck, 19 pelvis and 10 pancreas] and linear accelerators [DHX (n = 14) and Trilogy (n = 30)]. Moreover, the agreement dependency on the section of the phantom was assessed. Plan evaluations obtained by 2D, 3D, and volumetric γ‐index (both local and global) were also compared. Generally, high dose gradient resulted critically managed by the assembly, with a smoother effect in RA technique. Worse agreements emerged in the 2D γ‐index vs those of 3D and volumetric (P < 0.001), that were instead statistically comparable in global metric (P > 0.300). Volumetric plan evaluation was coherent with the average of passing rates on the 3 phantom axes (r ≥ 0.9), but transversal section provided best agreements vs sagittal and coronal ones (P < 0.050). The three studied districts furnished comparable results (P > 0.050) while the two LINACs provided different agreements (P < 0.005). The study pointed out that the phantom transversal section better fits the planned dose distribution, so this should be accounted when a two‐dimensional evaluation is needed. Moreover, the major reliability of the 3D metric with respect to the 2D one, as it better agrees with the dosimetric evaluation on the whole volume, suggests that it should be preferred in a two‐dimensional evaluation. Better agreements, obtained with RA vs IMRT technique, confirm that Octavius®‐4D is specifically conceived for rotational delivery. Lastly, the assembly resulted sensitive to different technology.


| INTRODUCTION
A careful pretreatment check with a patient-related quality assurance protocol (QA) is a crucial step in the pretreatment process in radiation therapy, because of plan complexity and high sophisticated technology behind the radiation delivery. Several tools are available for a two dimensional evaluation, as well as gafchromic films, Electronic Portal Imager Device (EPID), arrays of diodes, (as MapCHECK, SunNuclear Corporation) or of ionization chambers, (as MatriXX, IBA Dosimetry, and 2D-Array Seven 29™, PTW). Most of them allow to sum multiple irradiations into a single dose plane, applying correction factors in order to take into account the directional dependence on the gantry angle. 1 However, because of the complex, highly conformal three-dimensional shape of treatment volumes, a full 3D dose matrix, with a volumetric evaluation of composite fields, is strongly preferable to a planar dose value map. [2][3][4] In conventional linear accelerators (LINACs), dosimetric distribution corresponding to the different positions of gantry can be obtained by portal imaging and specific software, by algorithms computing the inner dose distribution by interpolation of measurements, as EpiQA ™ (EPIdos), through the GLAaS algorithm. Other assemblies furnish dosimetric data corresponding to the gantry position using phantoms characterized by opportune geometries or rotation capability, as SunNuclear ArcCHECK ® , Scandidos Delta4 ® phantom, the IBA Compass, and PTW Octavius ® -4D. Many studies report a comparison of these devices, assessing their different responses as well as their ability in detecting intentional errors, [4][5][6] or they were used in the validation process of novel QA strategies. 7,8 The present study is focused on PTW Octavius ® -4D. Its main characteristic is that it furnishes a time-resolved dosimetric acquisition as it rotates together with gantry, and allows the reconstruction of a volumetric dose distribution. The γ-metric, the standard technique used to evaluate the agreement between planned and measured dose, 9 can be obtained not only in a two-dimensional way but also three-dimensionally, allowing the evaluation of the whole volumetric dose distribution.
Nowadays, PTW Octavius ® -4D is widely used in the QA process, and its performances were already evaluated also for stereotactic treatments, [10][11][12] with flattening filter free (FFF) technology 13 or for testing respiratory-gated VMAT delivery. 14 In spite of this, some aspects of its performances were only partially investigated. 1,15,16 This study aims to characterize the 2D-Array in Octavius ® -4D for both IMRT and VMAT delivery techniques. The relationship among the different kinds of γ-index metrics was investigated. The dependence on the phantom axis was assessed as well as the impact of different LINAC and the delivery modality (if IMRT or VMAT). Three different anatomic districts were considered [head and neck (H&N), pelvis, and pancreas] in order to evaluate device response to dose distribution with different modulations and heterogeneities.

2.A | Octavius ® 4D: general description
The 2D-Array together with Octavius ® -4D are widely described in the literature. 1,16 Briefly, the 2D-Array consisted in a matrix of 729 vented ionization chambers distant 10 mm from center to center, embedded in a 27 × 27 array. Each chamber has the cubic size of 5 × 5 × 5 mm 3 and the effective measuring point is 7.5 mm below the surface of the detector array. The array is inserted into a motorized cylindrical polystyrene phantom, (diameter and length 32 cm and 34.3 cm respectively). Its capability to rotate synchronously with the gantry, in terms of angle and rotation speed as in the real planned treatment, is made possible, thanks to an inclinometer that is set on the gantry and that is connected to a control unit that transfers the movement information to the phantom and acquires dosimetric data every 200 ms. The beam always hits the detector array in a perpendicular way because the same face of the detector follows the gantry, so no correction factors are required.
The detectors of the 2D-Array have a calibration certificate for the central chamber in absorbed dose to water, that is independent from the Octavius-4D phantom (German National Laboratory, PTB, Braunshweing). For all the other chambers of the 2D-Array, a calibration file provides correction factors respect to the central chamber. So it is possible to perform an intercalibration of the central chamber measuring the dose in the phantom with an ionization chamber inserted into a chamber plate, following the IAEA Technical Report Series 398 approach, and defining a correction factor (K user ). Otherwise, it is possible to obtain the cross-calibration using, as reference value, the expected dose provided by the TPS in the condition defined by vendors (corresponding to: field 10 × 10 cm 2 , 200 MU, DR 300 MU/min, gantry at 0°) and using the so called K cross , that is the ratio between the TPS value and

2.B | Octavius ® -4D commissioning
The commissioning of Octavius ® -4D was carried out with the Trilogy Varian linear accelerator. PDD were measured for field sizes ranging from 4 × 4 cm 2 to 26 × 26 cm 2 measured at 85 cm Source to Surface Distance (SSD). The Octavius ® -4D CT used for plan verification was the artificial one provided by vendor. However, in order to better model the phantom in the TPS, a CT scan was performed and the obtained averaged HU was reported into the TPS. Moreover, the correct distance from the couch was measured and reported in the artificial CT by fusion of images. Static delivery tests included: 5 × 5, 10 × 10, 5 × 20 cm 2 static fields, a "pyramid" field-in-field shape given by superimposition of 5 × 5, 10 × 10, 20 × 20 cm 2 static fields. The evaluation of arc delivery performance was obtained using a 5 × 5 cm 2 arc and the combination of 5 × 5 and 10 × 10 cm 2 arcs, with a clockwise 180°rotation for both arcs, which were sequentially delivered in order to cover a full rotation of the phantom. The effect of different spatial directions was assessed by considering transversal, coronal, and sagittal planes and verifying whether there was a different response or not. Dosimetric data were assessed by the γ-analysis with acceptance criteria 3%/3 mm. 1,4 In order to achieve absolute dose, the central chamber of the array was cross-calibrated with a PTW Semiflex ionization chamber (volume 0.125 cc, type 31010), inserted into a RW3 slab replacing the 2D-Array inside Octavius ® -4D. The position of the chamber with respect to the isocenter was preliminarily verified by orthogonal kV images obtained by mean of the Varian On-Board Imager (OBI).
Measurements at the reference condition were carried out (field 10 × 10 cm 2 , gantry at 0°, 200 MU corresponding to 1.248 Gy for Trilogy and 1.344 Gy for Clinac 2100 DHX, i.e. the expected value in each measure session for the clinical plan verification) and the dose value was deduced from the chamber signal following the IAEA Technical Report Series 398 approach, taking into account also the correction for the daily LINAC output factor. The comparison with the central chamber measurement at the same conditions allowed deducing the so-called K user factor for each LINAC, useful for a validation of Octavius ® -4D for absolute dose assessment. This result was compared with the K cross , in order to evaluate the consistency of the two approaches (see Section 2.A).
Finally, it was evaluated that the dose distribution obtained by two uninterrupted clinical arc deliveries vs the same arcs delivered with up to four interruptions in order to understand the performance of the LINAC when an undesirable arc interruption happens and to test the inclinometer reliability.  Table 1, while each treatment plan is detailed in Supporting Information -Table S1.

2.C | Treatment plans
The dosimetric verification was carried out comparing the measured plan with the so-called "verification plan", where the dose distribution of the treatment plan was recalculated on the CT of the phantom performed with the in-home CT scan (the same used for commissioning), with a slice thickness of 3 mm. The TPS grid was set at 2.5 mm.

2.D | Evaluation of dose distribution
2D, 3D and the volumetric γ were evaluated with 3%/3 mm acceptance criterion, 4 because of its prevalent use in clinical practice, 17 and DD in local γ analysis was increased to 5% for doses lower than 0.1 Gy. Results by both local and global γ analysis definitions were investigated. The Γ < 1 was required to be satisfied at least in 95% of points. The analysis of results has been carried out by no ROI selection, as the whole volume was considered with a cut-off threshold set to 5% with respect to the maximum dose. This choice was coherent with threshold reported by an American survey of Nelms and Simon 18 (as more than 70% of 139 institutions involved in the study used a threshold between 0% and 10%).

2.E | Statistical analysis
Statistical analysis was carried out by the software package SPSS.20 (SPSS, Inc., Chicago, IL; USA). The gaussian distribution of samples was tested by Kolmogoroff-Smirmoff test after log e -transformation to reduce heteroscedasticity. Analysis of variance was studied for comparison of different groups and t-paired test for paired data (respectively, Kruskal-Wallis test for independent groups or Wilcoxon test for paired data if not parametric test was required). The significance level was set at P < 0.05.

3.A | Octavius ® -4D commissioning
For the Octavius ® -4D preliminary commissioning, dose profile for static 10 × 10 cm is reported in Fig. 1(a). The low-dose threshold was initially set at 0, but this revealed that the gamma index computation by the VeriSoft algorithm took into account also the boundary layer of the phantom, where no detectors were present [see Fig. 1(b)]. It was found that a 0.1% of the maximum dose was a sufficient low-dose threshold for the exclusion of this inconsistent area of comparison. Then the cut-off threshold was set at 5% of maximum dose, obtaining the change in agreement showed in Fig. 1(c), where the other delivery characteristics are unmodified.
The picture shows that the boundary layer appeared in good agreement, as it was not included in comparison, while the gradient area remained with scarce agreement.
The accordance among measured and calculated fields gave a percentage of passing points with global volumetric γ-index ranging from 95% to 100% for both delivered static and rotational fields.

3.B | Dose distribution verification
Pretreatment plan verification by mean of Octavius ® -4D resulted in a percentage of at least 90% passed points for 75.0% of cases for local volumetric γ-index (average value = 91.5 ± 4.1%) that rose up to 100% for global volumetric γ-index (average value = 97.9 ± 1.8%).

3.B.1 | Dependence on the γ-metric
The choice of the γ-metric implied different results in plan evaluation. The results evaluated on the three axes were resumed in their mean value in order to take into account the three dimensions together.
The distribution of the volumetric local and global γ-index and the corresponding 2D and 3D γ-index averaged on the three axes (transversal, sagittal, and coronal) are showed in Fig. 3, and their averages in Table 2.
T A B L E 1 Description of main treatment plan characteristics, grouped for studied anatomical region.

3.B.2 | Dependence on the phantom axes
Results of agreement in dosimetric distribution depending on the section of the phantom, if transversal, sagittal, or coronal, were evaluated and results are reported in Table 3.
The ANOVA test gave a statistically significant difference among the three sections for local γ-index, both 2D and 3D (P < 0.001).
The Bonferroni test showed that the difference was due to the higher value in the transversal direction with respect to the others (P ≤ 0.001). For the global γ-index, the difference was less significant (P = 0.034 and P = 0.063 for 2D and 3D respectively). γ-statistic of the three sections was always correlated with the volumetric γ-index (R ≥ 0.9, P < 0.001). If global γ-index was considered, the γ-index

3.B.5 | Dependence on the treatment region
The average values of γ-index for the three studied districts are summarized in Table 5. Even if the γ-index average values resulted always lower for head and neck treatments, the differences showed a not sig-

| DISCUSSION AN D CONCLUSION
Results underline that several factors affect plan evaluation when using Octavius ® -4D, and they are especially enhanced if the more restrictive local γ-index computation approach is used. Indeed, the global γ-index produces more homogeneous results with higher passing rates (Fig. 3), because its tolerance level is computed with    respect to the value of maximum dose. The first important issue is related to how the phantom is conceived, as it allows to consider different typology of metrics for the γ-index. The 2D approach considers each slice as independent of the surrounding volume, with the drawback that results are strongly dependent on the chosen plane, without a certain significant correlation between the magnitude of errors of different plans. 19 Such an aspect is then an undesirable characteristic of the 2D γ. The 3D analysis allows a slice-by-slice evaluation taking into account also the neighboring slices. The "extra" 3rd dimension used to search for agreement leads to a lower 3D γ-index with respect to the 2D, producing higher passing rates for patient QA if the same acceptance criteria are chosen. 20,21 Finally, the assessment of the entire volume under study, with a volumetric γ evaluation, is probably the main strength of this kind of phantom, especially if considering that it is exposed to the radiation Results also indicated a dependence on the section where the plan was evaluated: indeed the transversal section was always linked to a better agreement if compared with the coronal and the sagittal ones (P ≤ 0.001) and it gave global γ-index comparable with the 2D volumetric passing rates (P = 0.192; Poor agreement, obtained with static field at the commissioning step, is linked to the overly smoothened dose reconstruction due to the combined effect of the detector resolution and the interpolation performed by the algorithm [Fig. 1(c)]. This behavior was already underlined by Allagaier et al. 1 Table 4(b) (P = 0.019).
No dependence on the treated regions was found, as the difference in γ-index was not statistically significant (P > 0.065; Table 5). difference for both the LINACs vs 1-2% in Stathakis et al.) 16 , probably due to array-related characteristics. However, it was found that the K user was comparable with the K cross , obtained by the comparison with the TPS expected value (Δ ≤ 1%). The consistency of the two approaches suggests that the cross-calibration, necessary to take into account the daily output variation of the LINAC, could be performed by using the K cross approach, which is simpler than the K user approach.
In conclusion, the study pointed out that Octavius ® -4D is a very reliable tool, especially for VMAT pretreatment quality assurance, as very good agreements of treatment plans delivered with RA technique were found. The system resulted sensitive enough to catch differences in LINAC technology while it similarly managed pancreas, pelvis, and H&N treatments. A useful finding was that, in a twodimensional evaluation, the study showed that the transversal section better fits the planned dose distribution. Moreover, 3D slice evaluation is generally comparable with volumetric evaluation, suggesting that it should be preferred to the 2D when volumetric metric is not available.