Results of a two-year quality control program for a helical tomotherapy unit

Radiotherapy and Oncology 86 (2008) 231–241 www.thegreenjournal.com

Tomotherapy

Results of a two-year quality control program for a helical tomotherapy unit Sara Broggia,*, Giovanni Mauro Cattaneoa, Silvia Molinellia, Eleonora Maggiullia, Antonella Del Vecchioa, Barbara Longobardia, Lucia Pernaa, Ferruccio Faziob,c, Riccardo Calandrinoa a

Medical Physics, and bRadiation Oncology, Scientific Institute San Raffaele, Milan, Italy, cIBFM-CNR, Milan, Italy

Abstract Background and purpose: Image-guided helical tomotherapy (HT) is a new modality for delivering intensity modulated radiation therapy (IMRT) with helical irradiation: the slip ring continuously rotates while the couch moves into the bore. The radiation source (Linac, 6 MV) is collimated into a fan beam and modulated by means of a binary multileaf collimator (MLC). A xenon detector array, opposite the radiation source, allows a megavoltage-CT (MVCT) acquisition of patient images for set-up verification. The aim of this paper is to report the results of a two-year quality control (QC) program for the physical and dosimetric characterization of an HT unit installed at our Institute and clinically activated in November 2004, in order to monitor and verify the stability and the reliability of this promising radiation treatment unit. Materials and methods: Conventional Linac acceptance protocols (ATP) and QC protocols were adapted to HT with the addition of specific items reflecting important differences between the two irradiation modalities. QC tests can be summarized as: (a) mechanical and geometrical characterization of the system’s components: evaluation of alignment among radiation source–gantry rotation plan–jaws–MLC–MVCT; (b) treatment beam configuration in static condition: depth dose curves (PDD) and profiles, output factors, output reproducibility and linearity; (c) dynamic component characterization: accuracy and reproducibility of MLC positioning; rotational output reproducibility and linearity, leaf latency, couch movement constancy; (d) gantry–couch and MLC–gantry synchronization; and (e) MVCT image quality. Peculiar periodicity specific tolerance and action levels were defined. Ionization chambers (Exradin A1SL 0.056 cc), films (XOmat-V/EDR2), water and solid water phantoms were used to perform quality assurance measurements. Results: Over a two-year period the final average output variation after possible beam output adjustment was 0.2 ± 1% for the static condition and equal to 0 ± 1% for the rotational condition: around 98% of the collected output data was within the action level compared to 94% if no beam output adjustment was considered. An average energy variation of 0.4 ± 0.4% was found. The daily absolute dose verification of IMRT plans showed a dose reproducibility of 0.5 ± 1.2% and 0.4 ± 2.2%, for low and high dose gradient regions, respectively. Source–jaws–MLC and MVCT alignment results and jaw and leaf positioning accuracy were 6±1 mm. Couch–gantry–MLC synchrony tests showed good stability level (6±2 mm). Conclusions: QC results indicated good reproducibility of all HT mechanical–dosimetric performance. c 2007 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 86 (2008) 231–241.



Keywords: Helical tomotherapy; Quality control program

Helical tomotherapy (HT) is a modality for the delivery of inverse planned intensity modulated radiation therapy (IMRT), together with a highly integrated image-guided approach. It combines the main characteristics of a linear accelerator and a CT scanner [5,12,15,16,28]. Most commercial IMRT systems evolved from conventional linear accelerators equipped with multileaf collimators (MLC), whereas helical tomotherapy was specifically designed as an IMRT machine with many unique features. A linear accelerator is mounted on a ring gantry that continuously rotates



while the patient is translated along the gantry rotation axis during treatment delivery. The fan beam is subdivided into beamlets by means of a 64 leaf binary collimator, which provides a time modulation of the treatment beam. The same radiation source, combined with a detector array, allows the acquisition of 3D megavoltage images [18,21,22]. Thanks to the capability of reducing uncertainty in patient set-up and the ability to produce extremely high dose gradients and rapid dose fall off outside the target, this unit can provide better dose conformation around the target vol-

0167-8140/$ - see front matter c 2007 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.radonc.2007.11.005

232

Quality control program for a helical tomotherapy unit

ume, together with a corresponding dose reduction to normal tissue. A specific accuracy verification program for these highly complex plans must be developed to ensure confidence in clinical treatment delivery. The definition of a quality control (QC) protocol for an HT unit presents quite different challenges from those related to conventional Linacs [1,9,10,14,25]. Some similarities may be found with the sequential tomotherapy technique [27], where the main sources of treatment accuracy can be correlated to the proper alignment of the MIMiC collimator (Multivane Intensity Modulating Collimator) relative to the radiation beam and the collimator rotation, and to the correct and precise indexing of the patient table. Several papers in the literature report the definition of specific tests for a correct and complete helical tomotherapy delivery characterization. Balog et al. have proposed specific beam alignment [2] and dosimetric tests [3] for the characterization of the helical tomotherapy unit. More recently the same authors [4] proposed and developed a daily and monthly QC program in order to test most of the inherent features of the tomotherapy device, including megavoltage image quality, spatial and temporal resolution accuracy of the dynamic delivery properties and more traditional beam output characteristics. Fenwick et al. [11], in addition to providing a detailed description of the design and components of the machine, developed a QC program with experimental details for its implementation. However, no results and considerations regarding the stability and the accuracy of the system’s performance were reported. The output and energy stability of the HT unit has been described only by Mahan et al. [17] on a daily basis over a period of 20 weeks. Recently, transversal beam profile constancy was investigated by Langen et al. [13] over a seven week period, but the aim of their work was to investigate the use of a commercial diode array for the cone shape monitoring. The aim of this paper is to report the results of a twoyear QC program for the physical and dosimetric characterization of an HT unit installed at our Institute and clinically activated in November 2004, in order to monitor and verify the stability and the reliability of this promising radiation treatment unit. A brief description of the QC program implemented in our department, along with technical details concerning the implementation procedure, is also provided. This paper will focus on the treatment machine, not considering any control correlated with the integrated treatment planning system and with the patient quality control.

Materials and methods Helical tomotherapy unit characterization Helical tomotherapy (Hi-Art 2, TomoTherapy, Inc.) is a modality for delivering intensity modulated radiation therapy (IMRT) based on a helical irradiation pattern, obtained with a continuous rotation of the gantry while the couch translates through the bore of the machine. The machine uses a 6 MV Linac, without flattening filter, mounted on a

ring gantry at a source-axis distance (SAD) of 85 cm. The longitudinal field width (slice thickness) is defined by a pair of moveable jaws, able to define a maximum open field size equal to 40 cm · 5 cm. For patient treatments, the desired field size has to be fixed to one of the commissioned field sizes (in general approximately 1, 2.5 and 5 cm; in our case only 2.5 and 5 cm). In the lateral direction, the beam is modulated by a binary MLC, with a leaf isocentre width of 6.25 mm. Intensity modulation is accomplished by varying the aperture time of each leaf; the modulation pattern can change with the gantry angle and is defined over the course of a ‘‘projection’’, which corresponds to a gantry rotation of just 7, giving a total of 51 projections per rotation [5,12,15,16,28]. The unit not only represents a new IMRT modality, but is also a completely integrated imageguided system [IGRT]. The same radiation source, detuned to a nominal energy of 3.5 MeV, is also used for megavoltage image acquisition. A 738 xenon detector array mounted on the rotating slip ring, opposed to the radiation source, allows the 3D tomographic reconstruction of body structures in order to check and, if appropriate, correct, patient setup prior to treatment irradiation [18,21,22]. The HT unit is able to deliver highly dynamic treatments, the accuracy of which depends on the overall mechanical, geometrical and dosimetric performance of all the components. A complete and exhaustive characterization of the system unit should consider various aspects: a mechanical, geometrical and dosimetric description of all system components, both in static and rotational conditions; an accurate verification of the synchronization of the dynamic components and, finally, a description of the imaging system. The mechanical and geometrical characterization of a tomotherapy unit, as for a conventional Linac, deals basically with the alignment of all of the system’s components: lasers, source, moveable jaws, multileaf collimator (MLC), detector array. The impact of these parameters is similar to that of a conventional Linac with some specific differences, mainly related to the accuracy of laser alignment. The room in which the tomotherapy unit is located is equipped with seven lasers: two fixed green lasers able to define a ‘‘Virtual Isocentre’’ set outside the bore at a fixed distance of 70 cm from the ‘‘Real’’ radiation isocentre, and five moveable red lasers used to define patient set-up prior to treatment. Accuracy of the Virtual Isocentre position relative to the radiation isocentre inside the bore, together with the correct movement of the red lasers in relation to the fixed green ones, are fundamental prerequisites for correct and precise patient positioning and treatment. The dosimetric beam configuration includes traditional measurements such as beam output, energy, lateral and longitudinal profiles, output reproducibility and linearity, but with some small differences. The output calibration is defined, differently from a conventional Linear Accelerator, in terms of a reference dose-rate (cGy/min) at one reference geometrical condition and not based on a defined monitor unit (MU) number, irradiation time being the primary monitor of a tomotherapy unit. The main difference, when compared to a conventional Linac, is the cone shape of the lateral profile, due to the absence of the flattening filter.

S. Broggi et al. / Radiotherapy and Oncology 86 (2008) 231–241

Classical quality index of transversal dose profiles such as flatness and symmetry is not common for this type of system. On the other hand, the helical nature of the irradiation delivery renders the periodical acquisition and monitoring of the longitudinal profiles crucial, in order to avoid unplanned treatment overdose or underdose situations. The most distinctive feature of the tomotherapy unit is the capability of providing IMRT helical irradiation patterns, thanks to the concomitant and synchronized continuous gantry rotation, couch translation and opening/closing MLC leaf movement. For this reason, a precise characterization of the synchronized components is necessary, together with an accurate check of couch speed and couch movement. With the HT system, the patient is positioned outside the gantry bore; after MVCT–KVCT registration the couch automatically shifts to correct for possible set-up errors. Once the positioning procedure is terminated, for dose delivery, the couch translates from the virtual to the real radiation isocentre inside the bore and then translates continuously during irradiation along the total length of the target. HT, a highly integrated image-guided system, allows the acquisition of 3D CT images using the same treatment radiation source. A more accurate image quality assurance program could be useful in cases where not only bone structures but also soft tissues are to be considered during the KVCT–MVCT matching.

QC program implementation In Table 1 the current QC protocol for the tomotherapy unit implemented at our Hospital was summarized with the corresponding tolerance (TL) and action levels (AL). The implemented QC protocol was designed to systematically assess the parameters reported in the previous session. Conventional Linac QC protocols were adapted to the helical tomotherapy unit with additional items reflecting important differences between the two irradiation modalities. A demanding QC program was initially defined to better understand the mechanical and dosimetric performance of this new machine; based on the reported results, different periodicity and different tests can be proposed. Kodak XOmat-V films, scanned with a Vidar VXR16 film digitizer, and on-board MVCT imaging detectors were used for mechanical and geometric tests. For relative and absolute dose measurements, Standard Imaging A1SL ion chambers were used (volume 0.056 cc), calibrated in water and connected to two different electrometers: an eight channel Electrometer (TomoElectrometer) and a monochannel PTW Unidos. For absolute dose determination we followed the method proposed in the Thomas’s paper [24] to derive the quality conversion factor Kq for the Exradin A1SL ion chamber under helical tomotherapy reference conditions.

Daily checks Apart from a visual check of the fixed laser alignment with the reference position, daily QC tests were defined to verify the constancy and the reproducibility of the following dosimetric parameters: beam output in static and rotational conditions, beam energy, shape constancy of the cone profile and output reproducibility in a fully inten-

233

sity modulated treatment plan, in both low and high dose gradient regions. Output and energy constancy are measured for a static 40 cm · 5 cm maximum open field at gantry 0 with 70 s of beam-on time with the first 10 s delivered with all leaves closed to reach a stable output delivery. The dose is measured with a cylindrical ionization chamber (A1SL, Standard Imaging) placed at dmax (1.5 cm depth), 10 and 20 cm from the surface of a slab solid water phantom set at SSD = 85 cm. Energy constancy is defined by the ratio between the dose at 20 and 10 cm depths, both normalized to the maximum dose depth (1.5 cm), to take the output variation into account. Rotational output constancy and lateral profile shape are monitored through the acquisition, respectively, of the unit dose chamber signal and the MVCT detector array signal for a rotational irradiation with an open field of 40 cm · 1 cm. The daily output pulse signal and the daily cone shape are then compared with their reference counterparts defined during the system commissioning. To monitor the combined impact of jaw width, couch speed, leaf latency and MLC–gantry–couch synchrony, the dose is measured daily for a fully intensity modulated treatment plan, by using two ionization chambers (A1SL, Standard Imaging) inserted in a cylindrical homogeneous phantom (Cheese phantom). Two different measurements points are checked: one in a high dose/low dose gradient region and the other in a low dose/high dose gradient region. The total time for daily tests, performed by physics, is approximately 45 min, including phantom positioning, electrometer warm-up, chamber measurements and output and cone shape analysis.

Monthly checks In addition to the dosimetric tests already proposed for daily QC, the program of monthly checks aims at monitoring the correct alignment of the all system components: fixed and moveable lasers, the virtual isocentre position relative to the radiation isocentre, jaw twist, field aperture symmetry and gantry angle accuracy. To monitor these components the film tests and analysis proposed by Balog et al. [2] were implemented. To check the overhead cross-hair’s alignment relative to the radiation beam, a film positioned at the virtual isocentre and automatically moved to the radiation isocentre (nominal distance equal to 70 cm) is statically irradiated with an open field and the analysis of the performed image allows to check both if the laser is parallel to the radiation beam and also if the centre of the irradiated image (virtual isocentre) corresponds to the radiation beam centre. The moveable lasers should then correspond to the fixed lasers at the ‘‘home’’ position and exactly move based on predefined values with respect the fixed ones. The moveable collimating jaws must be aligned with the gantry rotation plane; a double exposure slit beam can be applied to a film placed at the isocentre, for gantry 0 and 180 with two different jaw settings. Two longitudinal profiles scanned from the field edge should match and the angle jaw/gantry plane can be related to the centre peak distance difference estimated by considering the 75% dose penumbra

234

Quality control program for a helical tomotherapy unit

Table 1 Quality assurance protocol for a tomotherapy unit Tolerance/action levels Daily checks (45 min) 1.1 Green laser alignment 1.2 Static output 1.3 Energy 1.4 Rotational output 1.5 Transversal profile constancy 1.6 IMRT plan constancy

LT: LT: LT: LT: LT: LT:

±2 mm ±1.5% ±2% ±1.5% ±2% ±2%/±4%

LA: LA: LA: LA: LA: LA:

±4 mm ±2% ±3% ±2% ±3% ±4%/±6%

Monthly checks (5 h) 2.1 Virtual isocentre alignment 2.2 Gantry position accuracy (0) 2.3 Jaw twist 2.4 Field centre constancy with jaw size 2.5 Field size accuracy: set field vs irradiated field for different gantry angle 2.6 Couch movement accuracy 2.7 Gantry–couch synchronization 2.8 Gantry–MLC synchronization 2.9 Static output 2.10 Energy 2.11 Rotational output 2.12 Transversal profile accuracy 2.13 Longitudinal profile accuracy 2.14 IMRT plan accuracy 2.15 Completion procedure check

LT: LT: LT: LT: LT: LT: LT: LT: LT: LT: LT: LT: LT: LT: LT:

±1 mm ±1 ±1 ±1 mm ±1 mm ±1 mm ±1 mm ±1 ±1.5% ±2% ±1.5% ±1.5% ±1.5% ±2%/±4% ±1.5%

LA: LA: LA: LA: LA: LA: LA: LA: LA: LA: LA: LA: LA: LA: LA:

±3 mm ±3 ±3 ±3 mm ±3 mm ±3 mm ±3 mm ±3 ±2% ±3% ±2% ±2% ±2% ±4%/±6% ±2%

Three-monthly checks (2 h) 3.1 MLC twist 3.2 MLC field size accuracy: set vs irradiated 3.3 o MLC–MVCT–gantry rotational plane alignment 3.4 MVCT image quality

LT: LT: LT: LT:

±1 ±1 mm; 2 channel 3 complete lines

LA: LA: LA: LA:

±3 ±3 mm 3 channel 2 lines

Annual checks (4 days) LINAC: mechanical/geometrical checks 4.1 Virtual isocentre alignment 4.2 Field divergence vs gantry plane 4.3 Jaw twist 4.4 Field centre constancy with jaw size 4.5 Gantry position accuracy (0) 4.6 Star shot of radiation isocentricity 4.7 Field size accuracy: set field vs irradiated field for different gantry angle 4.8 Couch movement accuracy 4.9 Gantry–couch synchronization 4.10 Gantry (static)–couch synchronization 4.11 Gantry–MLC synchronization 4.12 Couch speed uniformity

LT: LT: LT: LT: LT: LT: LT: LT: LT: LT: LT: LT:

±1 mm ±0.5 mm ±1 ±1 mm ±1 ±1 mm ±1 mm ±1 mm ±1 mm ±1 mm ±1 <2%

LA: LA: LA: LA: LA: LA: LA: LA: LA: LA: LA: LA:

±3 mm ±1 mm ±2 ±2 mm ±3 ±3 mm ±3 mm ±3 mm ±3 mm ±3 mm ±3 >3%

LINAC: dosimetric checks 4.13 Static outptut 4.14 Rotational output 4.15 Transversal/longitudinal profiles checks 4.16 Energy check: PDD curves 4.17 Rotational output reproducibility 4.18 Rotational output linearity vs irradiation time 4.19 Rotational output reproducibility vs gantry rotation period 4.20 Output reproducibility in dynamic condition (vs couch speed) 4.21 Output reproducibility for a simple IMRT plan 4.22 Completion procedure check

LT: LT: LT: LT: LT: LT: LT: LT: LT: LT:

±1.5% ±1.5% ±1% ±1% ±1% ±1% ±1% ±1% ±1% ±1.5%

LA: LA: LA: LA: LA: LA: LA: LA: LA: LA:

±2% ±2% ±2% ±2% ±2% ±2% ±2% ±2% ±2% ±2%

MLC: mechanical/geometrical checks 4.23: MLC–source alignment 4.24 Centre MLC–rotation gantry plane alignment 4.25 MLC twist 4.26 MLC field size accuracy: set vs irradiated for different gantry angle

LT: LT: LT: LT:

±1.5% ±1.5 mm ±1 ±2 mm

LA: LA: LA: LA:

±2% ±3 mm ±3 ±3 mm

MLC: dosimetric checks 4.27 Multileaf leakage 4.28 Leaf latency 4.29 Output factors

LT: <2% LT: <2% LT: ±1%

LA: >3% LA: >3% LA: ±2%

MVCT: mechanical/geometric checks 4.30 MLC–MVCT–gantry rotation plane alignment 4.31 Jaw–MVCT alignment

LT: 2 canali LT: ±1

LA: 3 canali LA: ±3

MVCT: dosimetric check 4.32 MVCT quality image 4.33 MVCT dose linearity

Daily, monthly, three-monthly and annual checks are reported with corresponding tolerance (LT) and action levels (AT).

S. Broggi et al. / Radiotherapy and Oncology 86 (2008) 231–241

levels, as 0.5 * a tan 1 [peak distance difference (mm)/profile distance (mm)]. To check the field aperture symmetry, a film set at the isocentre can be irradiated with different jaw setting and different opened and blocked leaves; longitudinal profiles taken across each irradiated field should be superimposed. As with conventional Linac, jaw width constancy for different gantry angles (gantry 0, 90, and 270) was assessed using film. A significant part of these tests is related to the synchronization and constancy check of all dynamic components: gantry, multileaf collimators and couch. To test synchronization performance, gantry–MLC and gantry–couch, the film tests proposed by Fenwick et al. [11] were implemented. Synchrony of leaf opening and gantry angle can be checked by placing two films axially on the couch that are irradiated opening two middle MLC leaves centered on three different gantry angles (0, 120, and 240); the delivery is correctly synchronized if both films show correctly angled star patterns. Gantry–couch synchronization can be verified by irradiating a film placed on the couch with leaf segments centred at the 0 gantry angle and opened at fixed rotations: three different segments spaced at a defined distance should be irradiated. In our protocol, the procedure to test gantry–couch synchronization is aimed at an amplification of possible non-synchronization effects: the peaks of the three irradiated segments lie 15 cm apart, along the direction of the couch drive, as opposed to the 5 cm suggested by Fenwick et al. [11]. Concerning dosimetric tests, the completion procedure generation is also verified; when the treatment’s irradiation interrupts there is in fact the possibility to generate a completion procedure in order to conclude the global treatment. The same IMRT plan is delivered with and without irradiation interruption, and the difference in dose between these two conditions is assessed in both low and high dose gradient regions. Total time required for all the tests included in the QC monthly program is approximately 5 h: three for the measurement and two for the relative analysis.

235

and two other opened lateral adjacent leaves and then at gantry 180 with the same lateral opened leaves (Fig. 1a); the distance (Fig. 1b) between the centre of lateral and central leaves should be constant upon gantry rotation. Linac, MLC and detector array should be aligned each other; the same detector should ‘‘look at’’ the same part of the beam as the gantry rotates. It can be tested by using the Tongue and Groove effect of the two central leaves, that can be revealed in the same detector channel for different gantry position. The measurement and analysis of these tests proposed for a three-monthly test could be performed in approximately 2 h.

Annual checks In addition to the mechanical, geometrical and dosimetric tests previously described, annual checks are conducted with the aim of verifying that beam configuration data, PDD curves, lateral and longitudinal profiles and leaf latency time match the commissioned values. As in the case of conventional Linac, beam output linearity and reproducibility is verified as a function of irradiation time and rotational period. The mechanical, geometrical and dosimetric tests defined in this annual QC program should check all the system

Three-monthly checks In accordance with our conventional Linac protocols, three-monthly tests were defined in order to better monitor MLC and megavoltage acquisition performance. The alignment of MLC–detector array–gantry rotational plane is assessed in combination with the accuracy of irradiated MLC field size and MVCT imaging quality. Also in this case, the film and detector array tests proposed by Balog et al. [2] were implemented. MLC is mounted independent from the moveable jaws and so MLC leaves could have a twist with respect to the plane of gantry rotation. The alignment can be checked with a film placed at the isocentre and double exposed with the same central opened leaf, at gantry angle 0 and 180. The resultant two images should be adjacent without any superimposition. The relative position of the MLC centre to the gantry centre of rotation can be determined irradiating a film first at gantry angle 0 with the opened central leaves (leaf 32, 33)

Fig. 1. MLC centre–gantry isocentre alignment: film dose image (a) and related transversal profile (b). The relative distance between the lateral and central Tongue and Groove effect denotes the position of the MLC centre relative to the gantry isocentre.

236

Quality control program for a helical tomotherapy unit

unit components (as in machine commissioning); the total necessary time could be estimated at approximately four days.

Results The results of a QC program over the period November 2004–December 2006 are reported in the following part of this work. Attention has been focused on our daily checks, due to both their very high number and the fact that they reflect most of the specific dosimetric aspects of the treatment unit.

Daily checks A quick visual check of the fixed virtual isocentre lasers always indicated good alignment, within 2 mm, relative to the reference position.

Over a period of about two years (November 2004–December 2006), with a total of 496 checks, globally an average static output variation of 0.1 ± 1.1% (1SD) was observed with a range of variation between 3% (December 2004) and +3.4% (June 2006); a similar average variation ( 0.2 ± 1.0%) was estimated by only considering the final measurements results in case of beam output adjustments with a range of variation between 3% (December 2004) and +2.3% (June 2006). For the latter collected data, around 97.5% are within the action level range (±2%), compared to 94% in case of no beam output adjustment. In Fig. 2 the daily percentage deviation estimated for static output is reported for all days in the considered period: Fig. 2a shows the data before any beam output adjustment, Fig. 2b the final data. No trend in time is clearly visible, with the exception of the first two months (November–December 2004), in which a lowering of physical output was observed. Excluding these first two months, in which no

Fig. 2. Daily percentage deviation for static output in the period November 2004–December 2006: (a) data before beam output adjustment and (b) final results after beam output adjustment.

S. Broggi et al. / Radiotherapy and Oncology 86 (2008) 231–241

beam output adjustment was implemented, an average output variation of 0.1 ± 1.0% (1SD) was estimated with a range of variation between 2.4% (November 2005) and +2.3% (March and June 2006), with 98.5% of the data within ±2% and 85% within ±1.5. Similar considerations can be made for the rotational output (427 data) with a global measured percentage deviation of 0.0 ± 1.0% (1SD) (range: 3.6% to +3.4%); similar mean deviation (0.0 ± 0.9%) was estimated by considering the final results after beam output adjustments with a reported range of variation between 2.4% (December 2004) and +2%. Excluding the first two months in which no beam output adjustment was performed, around 99.5% of the data was estimated within the action level ±2% and around 92.5% within ±1.5%. In Fig. 3, the daily percentage deviation estimated for rotational output is reported for the period November 2004–December 2006: Fig. 2a shows the global measured

237

data, Fig. 2b the final results after beam output adjustment. No data were collected between November 2005 and December 2005 due to problems in the analysis of the MVCT signal related to software upgrade incompatibility. A rotational output measurement with an ionization chamber should be introduced into our QC protocol to eliminate these kinds of problems. Good machine energy stability (479 data) (Fig. 4) was demonstrated with a mean percentage deviation equal to 0.4 ± 0.4% (1SD); of the collected data, 96.5% are within the 1% to +1% range and approximately 94.4% between 1% and 0.5%. Between February 2006 and August 2006 some spot deviations greater than 1.5% (although still within the action level) are clearly visible. We feel fairly confident that these results can be due to operator or measurement errors; this hypothesis is supported by the good concomitant results obtained the days before and after with no adjustment of machine parameters. A significant negative devia-

Fig. 3. Daily percentage deviation for rotational output in the period November 2004–December 2006: (a) data before beam output adjustment and (b) final results after beam output adjustment.

238

Quality control program for a helical tomotherapy unit

Fig. 4. Daily percentage deviation for energy in the period November 2004–December 2006.

tion is visible in the period November 2004–February 2006 ( 0.6 ± 0.2%) compared with a small positive deviation (0.1 ± 0.4%) in the period February 2006–December 2006, explicable in terms of a magnetron change (12th February 2006) and a different beam parameter’s setting. In Fig. 5 the cone shaped profile variation trend is reported, evaluating the average difference of all central detector channels. All the reported deviations are within the 0.9% to +0.8% range, if we exclude the deviation ( 6%) estimated at the beginning of November (2 November 2006), after a target replacement was scheduled. Observing Fig. 5, it is possible to recognize four regions qualitatively. A first region from November 2004 to June 2005 in which the cone shape shows a constant variation with a mean deviation of 0.0 ± 0.1% (1SD); a second region from July 2005 to November 2005 in which the cone shaped variation shows a slight decrease from +0.5% (July 2005) to 0.7% (Novem-

ber 2005), followed in February 2006 with the target replacement; a third region from February 2006 to June 2006 in which the cone shaped variation shows a constant mean deviation of 0.06 ± 0.21%. Finally, from July 2006 to the end of October 2006, it is possible to observe a slight decrease from +0.06% to 0.42% followed by a sudden average variation of approximately 6%, and then readjusted to the reference condition at the beginning of November 2006 with a second target replacement. In the beginning of July 2005, it is also possible to observe a trend variation from a constant null value to a positive variation around +0.5%, concomitant with a magnetron change. In Fig. 6, daily mean percentage deviations for the complete phantom IMRT treatment tests are reported for the high dose/low gradient dose region measurement point inside the target volume, as well as for the low dose/high dose gradient region measurement point. A larger dispersion

Fig. 5. Daily percentage deviation for transversal cone shape in the period November 2004–December 2006.

S. Broggi et al. / Radiotherapy and Oncology 86 (2008) 231–241

can be seen for the high dose gradient point, where the phantom positioning is more crucial. Good results were found for the absolute dose measurements (485 checks): mean percentage deviations equal to 0.5 ± 1.2% (1SD) (range: 5.9% to +3.3%) and to 0.4 ± 2.2% (1SD) (range: 10.6% to +9.9%) were measured for low and high dose gradients, respectively. Similar results were obtained by correcting the measured IMRT dose for the real daily rotational output. Action levels equal to ±4% and to ±6% for the two measurement points seem to be reasonable. Of the estimated differences, 97% are in fact within ±3% for the high dose/low gradient point and 93% of the deviations are within ±5% for the low dose/high gradient measurement point.

Monthly/three-monthly checks Fixed and moveable laser alignment showed very good stability; the virtual isocentre position and the moveable laser shifts were always within 1 mm, relative to the reference data. No correction of fixed and moveable lasers was ever necessary. Optimal geometrical and mechanical system performance was observed: the results of all proposed alignment and geometrical tests (source, jaws, MLC, and detector array) were always far smaller than the imposed tolerance levels of 1 mm or 1, respectively. Considering the dynamic components of the unit, the most critical parameter seems to be the gantry–couch synchronization. An average disagreement of 1.3 mm ± 0.6 mm was measured, slightly greater than the defined tolerance level (1 mm). The synchronization between gantry rotation and MLC movement was found to be within 1. Adequate periodical gantry rotation maintenance and gantry speed calibration seem to be very important issues. As concerns the dosimetric tests, the results are comparable with those obtained during daily checks. Good dose reproducibility is also found for the completion procedure verification. The mean percentage difference in the total

239

dose, measured with and without irradiation interruption, was equal to 0.4 ± 0.5% (1SD), both in the high dose/low gradient and low dose/high gradient regions, by considering different times of interruption. Quality image tests revealed no particular problems; the results are always within the tolerance levels and no array detector calibration was ever necessary.

Annual checks The principal aim of the annual QA protocol was to verify the stability of beam characteristics compared with that measured during commissioning. Measured PDD as well as lateral and longitudinal profiles matched the commissioned values within 1% in most points, and within 2% if the penumbra region of the profiles is considered. A slight difference was observed in the longitudinal field width and a consequent moveable jaw correction and calibration was carried out. Good geometrical and mechanical performance was demonstrated: all checks were, in fact, within 1 mm and 1.

Maintenance program In addition to a defined scheduled weekly and monthly maintenance program, a high level of technical intervention was required over this two-year period for the appropriate maintenance of our HT unit. Over 100 technical interventions were required for the following reasons: mechanical problems, dosimetric instability and a very few software/communications interruptions. Most of these problems were solved by our internal maintenance group, while in more critical situations an intervention on the part of tomotherapy was required. In addition to the replacement of crucial parts, two targets (10 February, 2005; 3 November, 2005) and one multileaf collimator (22 September, 2006), most of our machine problems have been related to an unstable beam pulse shape and to correct gantry rotation calibration. Concerning the beam output signal, a periodical afc (automatic frequency control) setting seems mandatory, concomitant with numerous

Fig. 6. Daily percentage deviations for a complete phantom IMRT treatment, for high dose/low gradient and low dose/high gradient measurements points.

240

Quality control program for a helical tomotherapy unit

pulse shaper changes (4 times) and magnetron replacements (22 November, 2005; 13 February, 2006; 13 March, 2006; 8 May, 2006; 8 September, 2006; 21 December, 2006). For correct gantry rotation calibration a periodical greasing and speed recalibration were required. Furthermore, several error messages are correlated to the couch movement; in our experience in only one case has the problem been effectively solved by means of the replacement of couch parts, and recalibration. Obviously parts of these technical interventions should be followed by an opportune QA program. We suggest that a target change should be checked by the mechanical and dosimetric tests able to verify the correct position: source–primary collimator alignment, beam centre–gantry rotation plane alignment, source–MLC alignment, cone shaped profile and topographic profiles’ constancy, energy and output check and finally some IMRT plans checks. The tests to be performed after a magnetron change could be: output and energy constancy, cone shaped profile constancy and IMRT plan verification. A MLC change should be checked by considering all the alignment tests relative to source, gantry rotation plane, radiation beam centre and array detectors. Based on these technical reports, the downtime for our tomotherapy unit was considerably greater than that expected for a conventional Linac accelerator. We estimate a downtime of around 9.3% for 2005 and around 11% for 2006, expressed as the ratio between the hours in which the machine does not work during the patients’ treatment time and the scheduled patients’ treatment time, considering 5.5 days a week and a daily patients’ treatment time of 8 h in 2005 and of 9.5 h in 2006.

Discussion and conclusions Several papers in the literature have attempted to define different tests and experimental methodologies to verify all the mechanical, geometrical and dosimetric system components [2–4,11,13,17,27], by presenting their sensitivity to possible beam and delivery variations. However, few of these papers have investigated the time constancy and stability for overall Hi-Art performance. In this work, the results of a two-year QC program for a tomotherapy unit system are reported. All the geometrical and mechanical tests showed very good time stability and constancy; moreover, defined tolerance and action levels, similar to the reference levels suggested by Fenwick et al. [11], seem consistent with the results found in our department. Geometrical and mechanical accuracy within 1 mm and within 1 was obtained, but no comparison could be made with other clinical experience. The optimum alignment observed among all the unit system components (source–moveable jaws–MLC) permitted a reduction in the frequency of these tests, e.g. three-monthly frequency for monthly tests 2.1–2.4. A wider comparison was possible with other clinical experience concerning dosimetric performance. Mahan et al. [17] investigated the output and energy stability of a tomotherapy unit over the initial 20 week period: differences within ±2% and ±1.5% were reported for static/ rotational output and for beam energy, respectively. In our analysis, higher stability and reproducibility variations were

observed, with discrepancies up to ±3% for static and rotational output, while beam energy showed similar behaviour. Based on our results, output stability seems slightly inferior compared with a conventional Linac: only 94% of the initial collected data were in fact within the defined action level, if no beam output adjustment was performed. Aside from a beam parameter’s setting and adjustment concomitant with system parts’ replacement (target, magnetron, pulse shaper), in our two-year experience the machine required a beam output tuning in approximately 20 days (around 4.5%), in order to report the measured output within the defined action level of ±2% [7,8]. Although the values reported in the literature for beam output constancy were relative to conventional Linacs that work in static conditions, we defined and maintained the same action level also for the helical irradiation modality of a tomotherapy system in order to satisfy the acceptance criteria reported and suggested in the literature for the global accuracy required in the dose delivery to a patient in a radiation treatment [6,19,20,26]. With regard to energy stability, the high level of reproducibility (1SD = 0.4%) demonstrated here should reduce the frequency of this test: it is felt that this parameter could be checked on a monthly basis, and every time after a beam output adjustment. The reported results seem to show neither any time trend variation nor any correlation with parts’ deterioration. As described in the results paragraph, the only correlation can be observed in Fig. 4, around the middle of February 2006 (12 February, 2006), where a different beam parameter’s setting concomitant with a magnetron change brought about a positive energy variation compared with the negative deviation estimated in the first period. No energy correlation was, however, observed for the other magnetron and/or target change. In agreement with the experience reported by Langen et al. [13], a slight decrease of the off-axis beam ratio was useful in diagnosing a thinning of the target, although the magnitude of the recorded decrease was less than that reported in the cited paper. As reported in Fig. 5, two target replacements have been scheduled in the course of our experience: the first one in February 2006 (9th) and the second one in November 2006 (3rd), both anticipated from a slight decrease in the cone shape variation in the two months before the change, respectively, estimated with a mean deviation equal to 0.53% (1SD = 0.13%) and 0.31% (1SD = 0.23%). However, the existence of a strict correlation between target degradation degree and decrease in the offaxis beam ratio seems not to have been demonstrated. To monitor the combined impact of all the system unit components, e.g. jaw width, couch speed, leaf latency, MLC–gantry–couch synchrony, the dose was checked daily for a fully intensity modulated treatment plan in two different measurement points: one in a high dose/low dose gradient region and the other in a low dose/high dose gradient region. Over the period studied, very good results were obtained, slightly better than those reported by Thomas et al. [23], where point dose measurements were performed for the verification of 10 patient treatments. The authors reported a mean percentage discrepancy for point dose measurements equal to 0.5 ± 1.1%, 2.4 ± 3.7% and 1.1 ± 7.3% for high dose, low dose and critical structure points, respectively. In our report comparable results

S. Broggi et al. / Radiotherapy and Oncology 86 (2008) 231–241

( 0.5 ± 1.2%) were found for the high dose/low gradient region, while better agreement and reproducibility were found for the low dose/high dose gradient region ( 0.4 ± 2.2%). All tolerance and action levels defined in our QC protocol were consistent with the results of periodical checks over a period of two years. Only the gantry–couch synchronization tolerance level was found to be too strict: a mean value of around 1.5 mm was in fact found, as opposed to the 1 mm defined as the tolerance level; this result could make the tolerance level equal to 2 mm, and the action level equal to 4 mm. Fenwick et al. [11] defined a lower action level (1 mm) for the same procedure; a possible explanation for this discrepancy can be found in the more extreme condition used in our protocol, where the overall couch drive distance is 30 cm, compared with 10 cm distance. In the two papers by Balog [4] and Langen [13] reported in the literature, great attention is devoted to the analysis of specific parameters and how their variation can be revealed by the measurements methods implemented. The aim of this study was simply to report the overall performance of the unit without proposing any particular measurement techniques. However, in agreement with Langen et al. [13], and based upon two years experience, we believe that certain independent measurement methods could be usefully introduced into our protocol, such as specific techniques for the measurement of the transversal cone shape (e.g. diode array), the rotational output (e.g. ionization chamber) based not solely on the internal monitor chamber signal and, above all, an independent measurement of the transversal cone shape. In conclusion, after two years of experience, thanks to the comprehensive QC protocol implemented, we can demonstrate and report optimal mechanical and geometrical performance of the system delivery; from a dosimetric point of view a high level of agreement was observed between calculated and measured dose distributions, although higher output stability could be desirable. * Corresponding author. Sara Broggi, Health Physics Department, Servizio di Fisica Sanitaria, IRCCS San Raffaele, via Olgettina 60, 20132 Milano, Italy. E-mail address: [email protected] Received 11 July 2007; received in revised form 31 October 2007; accepted 2 November 2007; Available online 3 December 2007

References [1] Aspradakis MM, Lambert GD, Steele A. Elements of commissioning step and shoot IMRT: delivery equipment and planning system issue posed by small segment dimensions and small monitor units. Med Dos 2005;30:233–42. [2] Balog J, Mackie TR, Pearson D, Hui S, Paliwal B, Jeraj R. Benchmarking beam alignment for a clinical helical tomotherapy device. Med Phys 2003;30:1118–27. [3] Balog J, Olivera G, Kapatoes J. Clinical helical tomotherapy commissioning dosimetry. Med Phys 2003;30:3097–106. [4] Balog J, Holmes T, Vaden R. Helical tomotherapy dynamic quality assurance. Med Phys 2006;33:3939–50. [5] Beavis AW. Is tomotherapy the future of IMRT? Br J Radiol 2004;7:285–95. [6] Brahame A, Chavaudra J, Landberg T, et al. Accuracy requirements and quality assurance of external beam therapy with photons and electrons. Acta Oncol 1988:27.

241

[7] Comprehensive QA for radiation oncology: report of AAPM Radiation Therapy Committee Task Group 40. Med Phys 1994;21:581–618. [8] Decreto Legislativo 19 agosto 2005, n187. [9] Esser M, de Langen M, Dirkx MLP, Heijmen BJM. Commissioning of a commercial available system for intensity-modulated radiotherapy dose delivery with dynamic multileaf collimation. Radiother Oncol 2001;60:215–24. [10] Ezzel GA, Galvin JM, Low DA, et al. 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; 30:2089–115. [11] Fenwick JD, Tome WA, Jaradat HA, et al. Quality assurance of a helical tomotherapy machine. Phys Med Biol 2004;49:2933–53. [12] Jeraj R, Mackie TR, Balog J, et al. Radiation characteristics of helical tomotherapy. Med Phys 2004;31:396–404. [13] Langen KM, Meeks SL, Poole DO, et al. Evaluation of a diode array (Tomo Dose) for QA measurements on a helical tomotherapy unit. Med Phys 2005;32:3424–30. [14] Low D. Quality assurance of intensity-modulated radiotherapy. Semin Radiat Oncol 2002;12:219–28. [15] Mackie TR, Holmes T, Swerdloff S, et al. Tomotherapy: a new concept for the delivery of dynamic conformal radiotherapy. Med Phys 1993;20:1709–19. [16] Mackie TR, Balog J, Ruchala K, et al. Tomotherapy. Semin Radiat Oncol 1999;9:108–17. [17] Mahan SL, Chase DJ, Ramsey CR. Technical note: output and energy fluctuations of the tomotherapy HiÆArt helical tomotherapy system. Med Phys 2004;31: 2119–20. [18] Meeks SL, Harmon JF, Langen KM, et al. Performance characterization of megavoltage computed tomography imaging on a helical tomotherapy unit. Med Phys 2005;32:2673–81. [19] Mijnheer B, Battermann J, Wambersie A. What degree of accuracy is required and can be achieved in photon and neutron therapy. Radiother Oncol 1987;8:237–52. [20] Quality assurance of treatment planning system-practical examples for non-imrt photons beams, booklet n 7, ESTRO, Brussels; 2004. [21] Ruchala KJ, Olivera GH, Schloesser EA, Mackie TR. Megavoltage CT on a tomotherapy system. Phys Med Biol 1999;44:2597–621. [22] Ruchala KJ, Olivera GH, Kapatoes JM, et al. MVCT image reconstruction during tomotherapy. Phys Med Biol 2000;45:3545–62. [23] Thomas SD, Mackenzie M, Field GC, et al. Patient specific treatment verifications for helical tomotherapy treatment plans. Med Phys 2005;32:3793–800. [24] Thomas SD, Mackenzie M, Rogers DW, et al. A Monte Carlo derived TG-51 equivalent calibration for helical tomotherapy. Med Phys 2005;32:1346–53. [25] Van Esch A, Huyskens DP, Bohsung J, et al. Acceptance tests and QA procedures for the clinical implementation of IMRT using inverse planning and the sliding window technique: experience from five radiotherapy departments. Radiother Oncol 2002;65:53–70. [26] Venselaar J, Welleweerd H, Mijnheer B. Tolerances for the accuracy of photon beam dose calculation of treatment planning systems. Radiother Oncol 2001;60:203–14. [27] Woo SY, Grant W, McGary JE, Teh BS, Butler EB. The evolution of quality assurance for intensity-modulated radiation therapy (IMRT): sequential tomotherapy. Int J Radiat Oncol Biol Phys 2003;56:274–86. [28] Yang JN, Mackie TR, Reckwerdt P, et al. An investigation of tomotherapy beam delivery. Med Phys 1997;24:425–36.