If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Development non gantry mounted ‘set-and-forget’ QA procedure for 9 beam parameters.
Freedom to measure at all gantry angles.
The sensitivity is comparable to the scintillator detector.
Measurement time is reduced by 40%.
QA results over 12 months at our pencil beam scanning proton centre.
To implement a single set-up monthly QA procedure for 9 different beam parameters at different gantry angles and evaluate its clinical implementation over a 12 month period.
We developed a QA procedure using an array detector (PTW Octavius 1500XDR) embedded in a rotational unit (PTW Octavius 4D) at our proton facility. With a single set-up we can monitor field central axis position, field symmetry, field size, flatness, penumbrae, output, spot size, spot position and range at different gantry angles (AAPM TG 224). The set-up is irradiated with homogenous 2D fields with dynamic aperture and spot patterns at five gantry angles. A modular top is used to check the range consistency. Absolute γ analysis were performed to compare measured dose distributions to calculated dose. All other parameters are directly extracted from the measurements. Additionally, the sensitivity of the set-up to small changes in beam parameters were compared to the Lynx detector (IBA).
Over a 12 month period, output, symmetry, and flatness were within ± 2 %; FWHM, spot positions, penumbra widths, and central axis fields were within ± 1 mm. Range differences were all within 1/2 of the energy spacing (±0.6 MeV) relative to baseline. Most (2 %, 2 mm) γ-analysis showed agreement scores higher than 90 %. The sensitivity is comparable to the Lynx detector and measurement time is reduced by 40 %.
The time-efficient monthly QA procedure that we developed can accurately be used to measure a large range of beam parameters at different gantry angles, within the TG 224 AAPM recommendations.
Proton therapy is a radiation treatment modality that uses proton beams to treat cancer. Over the past decade, several types of proton beam delivery systems (BDS) have been developed with increasing complexity. The proton BDS technology is expensive and to increase cost-efficiency long clinical treatment hours are typically implemented, leading to limited time to perform periodic QA measurements. Time efficiency is even more crucial for ad-hoc QA following corrective maintenance after machine downtime. The American association of physicists in medicine (AAPM) task group (TG) 224 provided comprehensive QA guidelines and methodologies for the commonly employed proton therapy treatment techniques.[
] The TG 224 report recommends performing beam parameter measurements (output, field symmetry, field flatness, range and spot size) on a monthly basis at different gantry angles relative to baseline. An ion chamber array detector is well suited to measure two-dimensional (2D) proton dose distributions.[
] Measuring these parameters with an array detector at different gantry angles can be difficult and labour-intensive, especially if the detector is not attached to the gantry. However, having a non-gantry mounted set-up ensures that the set-up is independent of the gantry system. The aim of this work is to propose a non-gantry mounted ‘set up and forget’ QA procedure that allows the measurement of a wide range of beam parameters at different gantry angles, without the need to change or align the set-up in between angles or different measured parameters. We describe how we implemented part of our monthly QA procedure and we show the results over a 12-month period. Additionally we looked into the sensitivity of the detector set-up and self-developed software for the analysis of all beam parameters.
Methods and materials
The set-up consists of an Octavius 1500XDR array detector (PTW Freiburg, Germany) inserted in the Octavius 4D phantom (Fig. 1). Controlled by an inclinometer attached to the gantry, the Octavius 4D phantom rotates synchronously with the gantry, keeping the detector array perpendicular to the beam. The Octavius 1500XDR consists of 1405 vented cubic ion chamber (IC) uniformly arranged on a 27×27 cm2 matrix. Each detector covers a cross-section of 4.4×4.4 mm2 with a height of 3 mm, resulting in an active volume of approximately 0.06 cm3. The chambers have a center-to-center distance of 7.1 mm. The device is positioned on the couch and aligned vertically using the room lasers. The phantom is levelled by adjusting the pitch and roll of the treatment couch. Kilovolt (kV) imaging is used to align the central ionization chamber of the array detector to the imaging isocentre (Fig. 1a/b) with 1 mm accuracy. 9 different beam parameters are determined at gantry angles 0, 45, 90, 135/160 and 180 degrees. For most beam parameters no phantom top is used (Fig. 1b). Only for the range consistency measurement a water equivalent modular phantom top with diameter 32 cm is added as shown in Fig. 1c.
Proton therapy at the Maastro clinic is delivered using a Mevion S250i Hyperscan system. The Hyperscan system consists of a small superconducting synchrocyclotron, attached to the gantry, that can deliver high energy proton beams using pencil beam scanning (PBS). The gantry system can rotate from 355○ to 185○. The treatment line is equipped with a BDS and an extendable nozzle at the end of the beam line. The nozzle of the device (Fig. 1) consists of the beam monitor (BM) system, a range modulation system (RMS) and dynamic field collimation system named adaptive aperture (AA). The RMS is used to obtain different energies and consists of 18 plates of different thickness. By combining these plates a total of 161 different energies can be obtained. The narrow pencil beams are magnetically scanned across the target volume by a single focus scanning magnet per isoenergetic slice. Additionally, the AA dynamically collimates each treatment field at each energy layer to sharpen the beam penumbra. More details on our beam delivery system are described by Vilches-Freixas et al.[
For the Octavius measurement set-up the extendable nozzle is placed at 18 cm from the isocentre for gantry angles 0 through 90, and at 33.6 cm for gantry angles 160 and 180 to minimise the air gap while avoiding collision with the phantom and/or treatment couch. This results in air gaps of 17.2 cm and 32.8 cm respectively, not taking the treatment couch into account for the latter. This cannot be further reduced due to the longitudinal edges of the phantom illustrated in Fig. 1b.
Treatment planning and beam parameters
The TG report of the AAPM recommends measuring output, field symmetry, field flatness, range and spot size at different gantry angles. Using this set-up the field central axis (CAX), field size, penumbrae and spot shape are additionally determined with the same measurement set-up. Table 1 gives an overview of the treatment plans used to determine the different beam parameters. The CAX, symmetry, field size, flatness, penumbra along the main axis and output are checked by irradiating the measurement set-up with a 2D 18×18 cm2 homogeneous field. These parameters are directly extracted from the 2D fields measurements using self-written Matlab code (Mathworks Inc., Natick, MA,USA). The phantom is positioned using orthogonal kV imaging, thus the CAX is a measure of the kV/beam coincidence. This parameter is determined absolutely and should be within ± 1 mm. Symmetry and Flatness are defined as stated in AAPM TG 24. Since symmetry values are so small we chose not to compare symmetry to a baseline value but to evaluate it absolutely. Reference values for field size, flatness, penumbrae and range are determined at time M0, i.e. yearly QA, by averaging over the gantry angles. Per parameter, two values are determined for beams not traversing the couch (gantry angles 0,45,90) and beams traversing the couch (gantry angles 135,160 and 180). This set-up is used to quantify the gantry dependency of the beam output. The absolute output verified daily, on the day of the monthly QA, with a parallel plane ionisation chamber at gantry 90 is used as a reference value.
Table 1Overview of treatments plans and corresponding beam parameters determined with those treatment plans. The absolute central axis position and field symmetry should be within ± 1 mm and ± 2 %. The reference values for field size, flatness, penumbrae determined at measurement time M0 are noted. The spot size is in turn compared to the spot shape predicted in the TPS using a gamma analysis. The field positions are compared to the expected spot position relative to the central spot depending on the spot spacing used in the treatment plan. And finally the reference L80 value was determined at measurement time M0.
Next, spot positions and spot sizes are checked by irradiating the measurement set-up with plans consisting of different spot patterns. A gaussian fit is performed for each measured spot and the offset with respect to the central spot position is determined with Matlab code. Spot positions are checked for energies 153.9 MeV and the pristine energy of 227 MeV. The combination of spot number, spot spacing and MU per plan that provided the minimal squared norm of the residual when performing individual spot Gaussian fits was selected. Additionally, the spot size and position were checked for energy 153.9 MeV by evaluating the dosimetric agreement between the measured and calculated dose using a 3D γ analysis implemented in Verisoft (v8.0 PTW Freiburg, Germany). A (global) γ criterion of 2 %/2 mm and a cut-off value of 30 % was used to optimize the amount of detector points used for the gamma analysis per spot. This means that voxels with doses below 30 % of the maximum calculated dose were ignored in the analysis. The reference treatment planning system (TPS) (Raysearch Laboratories, Stockholm, Sweden) dose was calculated using a Monte Carlo algorithm (Raystation 10A) on a calculation grid of 1 mm with an uncertainty of 0.5 %. Treatment plans were made on a synthetic solid cube set to water material. The water equivalent thickness (WET) of the effective measurement point of the array detector in combination with the phantom materials that the beam traverses was determined. Stelletjes et. al. determined that the inherent build-up of this detector has a WET of 8 mm.[
] The additional phantom material has a WET of 15.7 mm. Thus, a total WET of 23.7 mm is used as the reference depth where the TPS dose, calculated in water, is compared to the array measurement. For gantry angles 160 and 180 degrees the proton beam also traverses the treatment couch. The influence of the carbon fiber couch is taken into account by adding a WET of 30 mm to the reference depth when calculating the reference TPS dose for dose measurements at gantry angles 160 and 180 degrees.
Finally, to check the range at different gantry angles the standard phantom top is added to the Octavius set-up and irradiated with a homogeneous 2D 18×18 cm2 field (Fig. 1c). Due to the round surface of the semi-circular phantom top two peaks are measured in the detector plane as illustrated in Fig. 2a. The position of these peaks is sensitive to differences in range. Fig. 2b plots the transversal profile of three consecutive beam energies with an energy spacing of 2.1 mm. The peak-to-peak distance in the dose profiles is used to determine the range consistency. L80 is defined as the distance between the 80 % inner point of the two peaks as shown in Fig. 2b. ΔL80 is defined as the differences in L80 relative to the reference energy. Fig. 2b shows that a lower energy will lead to a smaller range and a larger L80 (positive ΔL80); whereas a higher energy will lead to a larger range and a smaller L80 (negative ΔL80). The extension is set to 33.6 cm for all gantry angles for the range measurement and this result in an air gap of 16.7 cm excluding the couch. When comparing the plans with two different nozzle extensions, we do see differences in lateral penumbra but not in the peaks position. Therefore, we chose a nozzle extension of 33.6 cm for all gantry angles to simplify the measurement setup while ensuring that we do not collide with the nozzle when rotating the 4D Octavius detector. All analysis are performed with self-written Matlab code. The reference L80 value is determined at measurement time M0 after yearly QA. Table 1 gives a summary of the measured beam parameters, the corresponding reference values and the applied tolerances.
All measurements are performed with the Octavius XDR1500 which has a detector spacing of 7.1 mm. As stated in Table 1, the tolerances limits recommended for spot position and range difference are ± 1 mm. To assess whether this setup was sensitive to these differences, despite the detector resolution, a few sensitivity checks were performed. The phantom set-up was irradiated by the described treatment plans and additionally by similar treatment plans where known small errors were introduced, i.e. known spot shifts and different beam parameters at gantry angle 0. The output sensitivity is checked by performing a BM linearity check. To assess the sensitivity to spot shape, the energy of the spots are changed by 10 MeV corresponding to a 10 % difference in spot sigma. Additionally the same adjusted treatment plans are also irradiated on a Lynx detector (IBA dosimetry) with an equivalent polymethyl methacrylate (PPMA) build up and measurement set-up. The Lynx detector consists of a scintillator screen coupled with a CCD camera in a light-tight box. The detector has an detector surface of 30×30 cm2 with a spatial resolution of 0.5 mm. The iris, variable aperture collimator system, regulated the amount of light reaching the camera. An iris opening of 30 % is used for all 2D field measurement and adjusted to 35 % and 60 % for the 227 MeV and 154 MeV spot plans respectively. Both Octavius and Lynx measurements are analysed with the self-written Matlab code. A tolerance limit of 1 mm is also recommended for the range differences. To check the sensitivity of the range check the difference in L80 distances are calculated for consecutive beam energies (1.1 MeV, the minimum energy step size).
The beam parameters are checked on a monthly basis using the described measurement set-up. Fig. 3 shows the trend analysis of all parameters extracted from 2D field measurements. The absolute field central axis position and symmetry values are plotted. Field size (defined as the Full width half maximum (FWHM)), flatness, penumbrae and output differences relative to baseline are shown. During the first few months, a gantry angle of 135 degrees was used to perform the measurements irradiating through the edge of the couch. After month 3 (M3) we switched to gantry angle 160 degrees to avoid irradiating through the edges of the couch. Field central axis position, FWHM and penumbra widths stay within ± 1 mm over all gantry angles throughout the year. Flatness values and symmetry values range in between ± 2 %. FHWM, flatness and penumbrae are out of tolerance on M7. On this date a wrong calibration file was used for the array detector. The calibration file ensure that the chamber to chamber variation of the response is less than ± 1 % when using the correct device specific calibration file. Output values stay within ± 1 % except for M3 and M6. As a result of these deviating results, maintenance was performed whereafter the output was within tolerance.
The difference in spot position is determined for × and y direction for each spot. Fig. 4 summarizes the difference of every spot relative to the expected spot position for all gantry angles and for the two different energies. Different measurement days are indicated by the different marker types and each spot position has a different colour. All deviations are within ± 1 mm throughout the year. The spread is smaller for the 25 spot/154 MeV plan and pristine energy, having the smallest spot sizes. Largest spread is seen at gantry angles 45 and 90. The spot sizes and positions of the 16 spots/227 MeV spot map plan are also compared to the measured dose with the TPS dose. All (2 %/2 mm) agreement scores are higher than 90 % from M4 onward (after optimisation spot patterns) except for dates M6 and M7. On these dates, deviations in the 2D field measurements were also observed.
To check the range at different gantry angles, the optimal energy of 151.7 MeV is determined experimentally. The differences in L80 distances relative to baseline determined after the yearly QA (M0), i.e. at gantry angle zero after the range is checked on the basis of Bragg peak measurements in a water tank, are shown in Fig. 2c. The baseline L80 value of 91.6 mm is determined for gantry angles 0–90 and 124.2 mm for gantry angles 160 and 180, taking the couch into account. The ΔL80 distance, L80 variations when comparing consecutive energies, are 17.5 mm for −1.1 MeV and 25.2 mm for 1.1 MeV differences with respect to 151.7 MeV. We defined the tolerance levels as follows: warning level at 6 mm (Δ0.4 MeV) and error level at 9 mm (Δ0.6 MeV) which corresponds to 1/3 and 1/2 of the ΔL80 of consecutive energies (1.1 MeV), respectively. All values are within the ± 9 mm tolerance except M7 where a wrong calibration file was used. Larger ranges are measured for gantry angles 160 and 180 (positive ΔL80 values) and smaller ranges for gantry angle 90 (positive ΔL80 values); these stayed consistent in time.
The sensitivity of the detectors for detecting the beam changes are quantitatively assessed by measuring treatment fields with known deviations. Table 2 shows the differences between the parameter determined with the reference QA plan and the adjusted plan per beam parameter. The same reference QA plan and adjusted plans are irradiated on the Lynx detector with equivalent build-up materials. The CAX sensitivity was determined by shifting the measurement set-up by 1.5 mm in both × and y direction. The CAX was determined using the matlab code before and after the shift. A difference in CAX of 1.5 mm is found in the horizontal and vertical profile, while the Lynx detector found a shift of 1.3 mm in the × direction and 1.7 mm in the y direction. Next, both detectors are irradiated with a treatment plan where a row of spots is added and deleted to the reference 2D square field. This should have an influence on both the symmetry and homogeneity of the fields in the vertical direction. A difference in symmetry of 2 % and a difference of flatness of 20.7 % and 19 % found by the Octavius and the Lynx respectively. The field size sensitivity is checked by removing the outer spots form the treatment field, resulting in a difference in FWHM of 3 mm. This FWHM difference is found by the Lynx and slightly overestimated by the Octavius 1500XDR detector. The penumbrae is changed by adjusting the margin used for the AA from 0.8 cm to 0 cm. Here the Octavius finds an average difference in penumbra width of 3.8 ± 1 mm and the Lynx 4.0 ± 1.7 mm.
Table 2This table summarises differences in measured beam parameters of adjusted treatment plans relative to reference treatment plan used during monthly QA procedure measured both with the 1500XDR OCTAVIUS detector and the Lynx detector. The figures illustrate the expected dose differences described in column 2 (Red/Orange indicates a higher dose than the reference plan, blue a lower dose). Results from 2D fields are shown per profile: H (horizontal profile), V (vertical profile), TL (diagonal from top left to bottom right), BL (diagonal from bottom left to top right). Penumbra values are determined both at left side (L) and right side (R) of the profile. Results of spot positions are averaged over all spots and in the case of spot positions averaged per group ‘’adjusted spots’’ or ‘’not adjusted spots.’’.
Adjustments to treatment plan (expected dose difference)
All spots shifted 1.5 mm in × and y direction
H: 1.5 mm V: 1.5 mm TL:0.0 mm BL: 1.4 mm
H:1.3 mm V: 1.7 mm TL: 0.2 mm BL: 1.6 mm
A row deleted, 1 row added and 2 rows deleted
H: 2.0 V: 0.0 TL:1.7 BL: 1.8
H:2.0 V: 0.3 TL: 1,8 BL: 1,7
Outer spots deleted.
H: 3.2 mm V: 3.1 mm TL:3.6 mm BL: 3.0 mm
H:3.0 mm V: 3.0 mm TL: 3.6 mm BL: 3.3 mm
A row deleted, 1 row added and 2 rows deleted.
H: 20.7 V: 2 TL:21.4 BL: 20.5
H:19.0 V: 1.3 TL: 19.3 BL: 20.1
Decrease lateral margin AA to 0 (from 0.8 cm).
HL: 4.5 mm HR:4.5 mm VL: 4.5 mm VR:4.1 mm
HL: 3.0 mm HR:5.6 mm VL: 5.6 mm VR:5.6 mm
Decrease energy spot pattern by 10 MeV
227 MeV: 80 % 154 MeV: 54.5 %
227 MeV: 88.7 % 154 MeV: 78.4 %
1 row spots shifted by 1 mm in × and y direction.
227 MeV: Shifted spots: X:1.0 ± 0.0 mm Y:1.0 ± 0.0 mm Non-shifted spots: X:0.0 ± 0.0 mm Y:0.0 ± 0.0 mm
154 MeV: Shifted spots: X:1.2 ± 0.4 mm Y:0.9 ± 0.0 mm Non-shifted spots: X:0.0 ± 0.0 mm Y:0.0 ± 0.0 mm
227 MeV: Shifted spots: X:1.0 ± 0.0 mm Y:0.9 ± 0.0 mm Non-shifted spots: X:0.0 ± 0.0 mm Y:0.1 ± 0.0 mm
154 MeV: Shifted spots: X:1.3 ± 0.4 mm Y:0.9 ± 0.0 mm Non-shifted spots: X:0.0 ± 0.0 mm Y:0.0 ± 0.0 mm
Since the spot sizes are evaluated using a (2 %,2mm) gamma analysis the decrease in agreement score is assessed when comparing a plan with a difference of 10 MeV to the QA treatment plan. It is shown that for the pristine energy plan this results in an agreement score of 80 % in the Octavius and 88.7 % in the Lynx, and the 154 MeV plan an agreement score of 54.5 % in the Octavius and 78.4 % in the Lynx. Finally, it is shown that a row of spots shifted by 1 mm can quantitively be determined using both detectors for both spot pattern plans.
All AAPM tolerances were met except for the symmetry measurement where the tolerance limit was expanded to ± 2 %. In the 2D array there is a spread in the response of the detectors. The detector is calibrated in a homogeneous cobalt beam leading to a homogeneity of ± 0.5 %. This could explain why the strict tolerance of ± 1 % (AAPM) is not met. The Octavius 1500 XDR detector can accurately be used to distinguish shifts in spot positioning and spot spacing down to 1 mm despite the detector spacing of 7.1 mm. The larger spread in spot positions on gantry angles 45 and 90 could be explained by the beam steering settings which are more critical and challenging at these angles. These angle dependent measurements triggered machine interventions on M3 and on M6, that would have not been noticed if we did measurements from only one gantry angle.
Proton range is a crucial beam parameter that needs to be checked periodically. In the case of our machine, only the highest energy of 227 MeV is extracted from the synchrocyclotron and it is further degraded at the RMS located at the nozzle to generate a total of 161 energies. According to Gomà et al. our BDS is of Type I because the energy selection system is located downstream the beam monitors.[
] Because the different energies are generated using combinations of range shifter plates, and not generated upstream at the accelerator level, measuring only one energy at five different gantry angles seems appropriate for a gantry dependent range consistency check of our system. Furthermore, on a daily basis we already check four different energies which are generated as a combination of these range shifter plates (Addendum A).
We quantitively compared the sensitivity of the Octavius 1500XDR with the Lynx detector for all beam parameters. All adjustments to the 2D fields where detected by both detectors. Although the 1500XDR has as lower resolution than the Lynx detector, this is sufficient to measure deviations in beam parameters within the relevant tolerance limits. Due to the lower amount of measurement points in the 1500 XDR detector, changes in shape will lead to smaller agreement scores more quickly. By determining the known deviations in the treatment plans with our self-written Matlab software, the software is inadvertently validated to correctly determine the deviations.
We developed a monthly QA procedure where different beam parameters can be measured at different gantry angles using a single measurement set-up. The phantom rotates the detector synchronously with the gantry and thus no adjustment of the set-up is needed for different gantry angles. By irradiating the measurement set-up with three types of treatment plans (Table 1) 9 beam parameters can be evaluated. The phantom is positioned using orthogonal kV imaging, adding an additional check of the kV/beam alignment. The 2D square fields are dynamically collimated using the AA. Thus, the accuracy of the AA is also checked with this procedure. The AA is calibrated to compensate for the lateral run-out of the nozzle per gantry angles. In contrast to most gantry mounted devices, typically used for gantry dependent measurements, the detector rotates independently from the nozzle, which could better pick up gravitational effects. The whole procedure can be done within 80 min, with about 20 min of setting up the device and 60 min irradiating all fields. This is a significant improvement from our previous QA set-up where we used 3 different gantry set-ups resulting in a measurement time of 140 min. Previously we measured these parameters with the Lynx detector in combination with the Sphynx phantom (IBA dosimetry). This limited the amount of gantry angles to only three orthogonal directions. With this set-up we have the freedom to perform measurements at all gantry angles. This monthly QA procedure that we developed can accurately be used to measure a large range of beam parameters at different gantry angles with a significant time gain of about 60 min. This set-up can also be implemented at other proton gantry facilities because the equipment is placed on the treatment couch. Given the pulsed nature of our beam, we chose for the PTW 1500XDR detector to avoid saturation effects, but other commercial 2D array detectors can be used in pulsed or continuous beam facilities. Further improvements include the integration of the monthly QA results in myQA software (IBA, Louvain-la-Neuve, Belgium) to improve the trend analysis. This measurement set-up is a part of a more comprehensive monthly QA program described in Addendum A. Beam monitor linearity check could be added to the beam parameters determined with the Octavius 4D set-up.
The monthly QA results that we obtained over a one year period are within AAPM TG 224 tolerances for 5 different gantry angles. The ‘set-and-forget’ QA procedure that we developed has several advantages:
The set-up is independent of the gantry system.
A single set-up is used for a large range of beam parameters.
The set-up gives the freedom to perform measurements at any gantry angle.
It results in a significant improvement in measurement time.
The sensitivity of the set-up to detect small changes in beam parameters is comparable to the Lynx detector (IBA).
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
The authors would like to acknowledge the physics innovation team at Maastro clinic, specifically Bas Nijsten, Ans Swinnen, Esther Kneepkens, Marcel Walpot, and Wiel Habets for performing and checking the monthly QA procedure.