A quality assurance program for ancillary high technology devices on a dual-energy accelerator

Transcription

1 ELSEVIER Radiotherapy and Oncology 38 (1996) 51-6 R ADIOTHERAPY aonco~~~~ A quality assurance program for ancillary high technology devices on a dual-energy accelerator Eric E. Klein*, Daniel A. Low, Derek Maag, James A. Purdy Mallinckrodt Institute of Radiology, Washington University School of Medicine, 51 S. Kingshighway, SI. Louis. MO 631 IO. USA Received 19 May 1995; revision received 23 August 1995; accepted 29 August 1995 Abstract Our facility has added high-technology ancillary devices to our dual-energy linear accelerator. After commissioning and acceptance testing of dual asymmetric jaws, dynamic wedge, portal imaging, and multileaf collimation (MLC), quality assurance programs were instituted. The programs were designed to be both periodic and patient specific when required. In addition, when dosimetric aspects were affected by these technologies, additional quality assurance checks were added. Positional accuracy checks (light and radiation) are done for both asymmetric jaws and MLC. Each patient MLC field is checked against the original simulation or digitally reconstructed radiographs. Off-axis factors and output checks are performed for asymmetric fields. Dynamic wedge transmission factors and profiles are checked periodically, and a patient diode check is performed for every new dynamic wedge portal. On-line imaging checks encompass safety checks along with periodic measurement of contrast and spatial resolution. The most important quality assurance activity is the annual review of proper operation and procedures for each device. Our programs have been successful in avoiding patient-related errors or device malfunctions. The programs are a team effort involving physicists, maintenance engineers, and therapists. Keywords: Quality assurance; Dual asymmetric jaws; Dynamic wedge; Portal imaging; Multileaf collimation 1. Iotroduction The American Association of Physicists in Medicine (AAPM) Task Group Report (#4) recently published [2] a comprehensive manuscript that touchs on every aspect of external beam radiation oncology quality assurance (QA), including machine QA, treatment planning, and patient treatment QA. Although the report addressed the need to develop QA programs for new high-technology devices placed on accelerators, it clearly stated that these devices were beyond the scope of the report. Dual asymmetric jaws, dynamic wedge (DW), an electronic portal imaging device (EPID) (Portal Vision, Varian Assoc., Palo Alto, CA), and multi-leaf collimation (MLC) (Mark II MLC, Varian Assoc., Palo Alto, CA) were added to our dual energy linear accelerator l Corresponding author, Tel.: , Fax: (C121OOC, Varian Assoc., Palo Alto, CA). These technologies were clinically implemented after rigorous acceptance testing and commissioning. Treatment and periodic QA procedures were established and implemented. Our machine is not yet equipped with a record-and-verify system (R&V). We have concluded that an R&V system is essential. This report describes in detail our QA program for the previously mentioned high-technology devices. During the accelerator s annual full calibration, the ancillary devices are checked to ensure maintenance of acceptance specifications and clinically used parameters. A subset of these measurements is performed either daily, weekly or monthly. Decisions on accuracy requirements mainly follow recommendations of TG 4 for related items (jaw accuracy, wedge factors, etc.). Some criteria are also based on clinical requirements for beam delivery. Commissioning details are referenced in this report. Other investigators have documented a quality assurance /96,% Elsevier Science Ireland Ltd. All rights reserved SSDI (95)1634-S

2 52 E. E. Klein et al. /Radiotherapy and Oncology 38 (19%) 51-6 program for computer-controlled radiotherapy [ 11. There also exists comprehensive documentation on QA of linacs [1,9]. 2. Methods and materials 2.1. Dual asymmetric jaws The Y-Jaws (upper) have the ability to independently sweep 1 cm beyond isocenter, while the X-jaws (lower) can independently sweep 2 cm past isocenter. The most prevalent position used by asymmetric fields is at. cm to create a nondivergent field edge. However, with the introduction of MLC, our use of asymmetric jaws has more than doubled, as the jaws must move as close as possible toward the shaped field and be used to cover regions of leaf abutment and trailing leaf edges. If the ability of the jaws to move asymmetrically is an upgrade to the machine, recommissioning of jaw position accuracy (symmetric and asymmetric) must take place. Rescanning of both photon and electron field profiles must be performed. The acceptance testing of the asymmetric aspect includes checks of asymmetric jaw position accuracy over the entire jaw range and for various gantry angles. Our clinic has chosen and confirmed a basic alteration of our monitor unit calculation algorithm to account for the fact that the asymmetric fields have a geometric center that is offset from the center of the flattening Elter [8]. Off-axis factors (OAFS) are used to correct the output, namely the collimator scatter factor component. We have measured OAFS in air along the principal axes and the diagonals. The OAFS were measured in air using a farmer ion chamber with brass buildup caps appropriate for the energy being measured (6 or 18 MV). The offaxis readings were normalized to the reading on central axis and tabulated as a function of the radius. The OAFS used for clinical calculations were averaged over all measured directions, as they differed from each other by no more than 1.2%. Another parameter that may be influenced by the ability of the jaws to move asymmetrically is backscattering of photons into the monitor chamber. This has been reported to influence the output for very elongated symmetric fields [6]. Because the Y jaws can sweep well past isocenter, the output should be checked under these conditions during acceptance testing. This problem has not been found with the newer Varian monitor chambers (Kapton type). On a monthly basis, each independent jaw is checked by comparing jaw setting vs. the light field position vs. 5% radiation value (edge) for fields designed as Fig. 1. Film of superimposed quadrant fields (1 x 1Ocm ) matched along the isocentric planes.

3 E.E. Klein et al. /Radiotherapy and Oncology 38 (1996) quadrants (two nondivergent edges). The jaw position accuracy has been specified with an accuracy of 1 mm for setting vs. light field at all positions, which exceeds TG 4 recommendations. A criterion of f.5 mm is applied for light vs. radiation field. Jaw positions are set, and the light field illuminated onto graph paper. The light field is then, etched onto film that is exposed to radiation. Silver Bromide film (XV Kodak, Rochester, NY) is used for the film dosimetry. A superimposed film (Fig. 1) is taken with each quadrant irradiating the film separately. Ideally the composite film should exhibit no distinct regions of overlapped regions or gaps. On a monthly basis, the simulator s asymmetric jaw (wire) settings accuracy is checked for the O.O- and lo.o-cm settings. An accuracy of 1 mm is expected, which is the same criterion as set for symmetric fields. Annually, the same quadrant test is performed at the four cardinal gantry angles. An isodensity scan of the superimposed film should show dose homogeneity across all intersections. A dose inhomogeneity of I f 5% (when compared with a point away from the junction) over a distance of I 2 mm is acceptable. These limits were felt to be significant, so that, if exceeded, clinical junction changes would be necessary. Annually, OAFS (used for asymmetric jaw monitor unit calculations) are spot checked to ensure agreement within.5% of the tabulated OAFS. A farmer-type ionization chamber (m N23333, CNMC Co., Nashville, TN) in combination with electrometer (Keithly 62, CNMC Co., Nashville, TN) is used with appropriate brass buildup caps. The output at d,, off-axis when the Y-jaw is 1. cm beyond isocenter should be checked annually to ensure that backscattering to the monitor chamber that can cause an eventual decrease in output is not occurring. This measurement (at the respective d,,, depths) is compared to calculations using the appropriate OAF. Annually, corresponding quadrant films are also taken on the simulator. A summary of the QA checks performed is found in Table Dynamic wedge The capability of driving a jaw (Y) while changing the dose rate has led to the use of that technology to produce angled isodoses that mimic wedged fields. This technology is called dynamic wedge and is governed by segmented treatment tables (SITS) that configure the jaw position in relationship to the portion of treatment delivered. [4] There are currently 256 SITS, one for each photon beam energy, wedge angle (15, 3, 45 or 69, and field size ranging from 4. to 2. cm in.5~cm steps. The SIT files are write-protected. PW 3 CLINAC 21oOC SN 138 MORNING CHECKOUT LOG 2/6/95 16:45 Rtt: Checked by: Table 1 Asymmetric jaws quality assurancea Frequency Test Tolerance Monthly Annually Jaw setting vs. light field vs. radiation field 4-quadrant exposure Simulator setting vs. wire position Jaw setting vs. light field vs. radiation field at cardinal gantry angles Off-axis factors along three main axes Water scans of beam split fields and asymmetric fields (5 cm from midline and IO cm over midline) Radiation output with Y-jaw 2 cm over midline Review of procedures and inservice with therapists 2mm Homogeneity within f 5% over 2 mm Imm Imm.5% 5% radiation edge within I mm of jaw setting 1% measured vs. calculated dose All operators must fully understand operation and procedures?3ee Ref. [8] for recommended commissioning and tests. DYNAMIC T&ATht!ZNT SUMMARY DATA TREATMENT SCREEN INDEX : 11 TREATMENT PARAMEIER SETITNGS : STATISTICS SUMMARY DOSE STD DEV (Mu ) :.2 POSITION STD DEV ( millimeters ) :.13 DOSE-WEIGHTED POSITION STD DEV ( mil$eters ) :.8 NUMBER OF SAMPLES S7T SUMMARY SETnNGS ACTUALS INSTANCE DOSE (MU) Y l(cm) DOSE (Mu) Y l(cm) : M) iii rn- loo.4-5. NOTES Fig. 2. STT summary for a 6-MV, 6, 2-cm dynamic wedge field. A statistical check includes a check sum analysis (dose-weighted position check).

4 54 E.E. Klein et al. /Radiotherapy and Oncology 38 (19%) 51-6 Inherent self checks are performed for dynamic wedge. A comparison SIT (anticipated vs. actual) is run for each treatment and for each anticipated jaw position and monitor unit interval. The comparison and check sum (Fig. 2) are calculated before (anticipated) and after (actual) treatment. For each SIT, a transmission wedge factor was measured to construct the large tables of wedge factors used for the monitor unit calculations. This is necessary because of the lack of a smooth dependence of wedge factors on the field size. Initial commissioning called for measurements of a subset of isodose profiles (approximately one-third of the SITS). Acceptance testing should confirm the ability to limit the availability of SITS to only those desired (i.e., currently in clinical use). Confirmation should also be made that dose profiles are independent of wedge direction (which jaw is set in motion). The ability of the SIT to be completed despite interruptions should be tested by terminating and restarting the beam during a dynamic wedge run. The commissioning process was described by Klein et al. [4]. Checks are performed that are patient specific for DW. Each time an SIT is designated for its first clinical use, the SIT is printed, and a film profile at dmax is exposed and scanned. This has led to a library of SITS with corresponding profiles. The profile film is intended as an efficient check to ensure the DW gradient is reasonably correct. The film is checked for density at three points (central axis and at 8% field intervals) with known separations. The corresponding density change from each 8% width point to central axis is averaged and divided by the distance (central axis to 8% width). The calculated value must correspond to the expected ratio to within.5% of the expected ratio. For example, 1.5% per cm for a 15 6 MV dynamic wedge is expected and a range of l-2% per cm would be acceptable. A measurement out of range would lead to isodose measurements for the particular SIT. For each dynamic wedge field a diode check is performed. The diode (Isorad, Nuclear Assoc., Carl Place, NY) is placed at central axis on the patients skin surface during the first or second fraction. The diode electrometer (Model 22D, CNMC Co., Nashville, TN) reading is corrected by factors that depend on SSD, field size, and diode response for that day. The corrected reading corresponds to the dose at dm. Deviations of 5% or less are considered acceptable due to the diode s systematic limitations in terms of spatial resolution and placement by the therapists. A second reading and investigation are performed for larger deviations. The technologists are also instructed to illuminate the light field at the end of each fraction to confirm that the remaining light field strip corresponds to the toe of the wedge. A logbook is maintained to store the SITS, profile films, clinical use information (treatment site, etc.), diode results, etc. On a daily basis, the SITS in clinical use are printed Table 2 Dynamic wedge quality assurances Frequency Test Tolerance Patient specific Daily Monthly Annually Scan of orthogonal film profile at 4, Printout of SIT First day central axis diode check Daily check of light field after treatment completion Check sum performed by controller system for each treatment Functional test and printout of clinically used SITS, with comparison to original data Check of active (commissioned) STTs Check of wedge factors (for 32 SITS covering range) Profiles for 16 STTs covering range Review of procedures and inservice with therapists See Ref. [4] for recommended commissioning. Profile gradient (% change/cm) within.5% of expected values No change from original 5% Confirm wedge direction Internal check No change Removal of inactive S-ITS 2% Wedge angle within 2 All operators must fully understand operation and procedures and checked. The user has the ability to limit only particular SITS available for use. We allow use of the required SITS as treatments require them. On a monthly basis, SITS no longer in use are given a non-available status. Spot-checks of DW factors are carried out in phantom (Solid Water) at a depth of 1. cm on an annual basis. A deviation of 12% (in accordance with TG 4 for physical wedges) is acceptable. Large field DW isodoses are measured with film. When they are compared to commissioned isodose profiles (used for treatment planning data), a deviation of 52 in wedge isodose angle is chosen for the tolerance. A summary of the QA checks performed is shown in Table Portal vision The EPID system consists of an imaging cassette, readout electronics, a control computer, and a display computer. The imaging cassette and electronics are mounted to the accelerator gantry through a support arm that swings roughly 9 between the retracted and

5 E.E. Klein et al. /Radiotherapy and Oncology 38 (19%) extended positions. The imager can be positioned from 17 cm target-to-imager distance (TID) to 15 cm using two-speed drive motors operated by switches placed on the support arm. A molded fiberglass piece covers the chamber and acts as both a safety shield and a collision sensor. A collision with the chamber cover (or a cover over the swing arm) causes an interlock condition that restricts gantry rotation and vertical couch movement. Certain electronic components within the cassette are sensitive to radiation and should not be exposed to the primary photon beam. Restrictions to the irradiated field sizes are therefore required and are a function of TID. The chamber operates through the ionization of a. l- cm thick layer of isooctane liquid placed between two circuit boards. There are 256 parallel conducting strips on the collection board that are connected to an electrometer, and their output is multiplexed to a digital-toanalog converter. The electrometers are scanned multiple times until roughly 1 or 2 ms have passed, depending on the chosen imager acquisition mode. The entire scan process takes roughly 5-1 s and is synchronized with the accelerator beam pulse repetition rate to ensure that the beam intensity will remain relatively constant with respect to the data acquisition scans. A full description of the commissioning process was described by Low et al. [7] Our QA process consists of five parts: (1) physical operation and safety; (2) image acquisition, resolution and noise; (3) image storage, analysis and handling; (4) reference image acquisition; and (5) operational techniques. The physical operation and safety tests relate to the motions of the unit, stability of the unit supports, safety interlocks, and interlock overrides. Because the EPID acquisition system is nonlinear, quoting the noise as a fraction of the signal has little quantitative significance. Imager sensitivity and noise characteristics are measured using a technique that predicts the thinnest detectable imbedded feature (e.g., bone) in a homogeneous phantom. The staff member who performs the daily accelerator QA procedure also conducts a weekly imager QA (Fig. 3). The couch and gantry motion interlocks are checked, as well as the operation of the imager arm lock when the gantry is not pointing downward (18 f 7 ) Spatial resolution is measured using a modified split-field technique and daily image QA is measured using a modified contrast-resolution phantom. The phantom is called the Las Vegas phantom (Fig. 4). The phantom is imaged with the standard acquisition mode of 24 MU/min at both 6 and 18 MV. The phantom is placed at isocenter, with a 13 x 13 cm* field size that just covers the phantom. A simple count of the visible holes is used to ensure that the image quality is consistent. Images taken during the commissioning procedure serve as landmarks for later checks. A checklist is completed that indicates the successful completion of each item as well as the number of holes visible on the phantom. The phantom provides a simple measurement of the spatial and contrast resolutions. Annually the imager calibration is performed. The procedure consists of a no-radiation image to compensate for signal offsets, as well as a flat-field image. The PORTAL VISION MORNING CHECKOUT SHEET Date: Initial: Locking Lever Interlock at Gantry Angle of *So from 18 Interlock Lamp Lights When Locking Lever is Unlocked at 1gO Casette Auto Drives to 4 cm Gantry/Couch Vet-t. Interlocked at Cassette Gantry/Couch Interlocked at S-Arm Touch Plate Enable Lamp Lights When Motion is Activated 6x-Test Pattern Resolution 7 Holes 18x-Test Pattern Resolution 7 Holes Cassette Stores Automatically at 7 cm in Standby Fig. 3. Portal vision weekly checkout sheet. Fig. 4. Portal image of Las Vegas (resolution) phantom that includes various combinations of diameters and depths for the, etched holes.

6 56 E.E. Klein et al. /Radiotherapy ana Oncology 38 (19%) 51-6 Table 3 On-line imaging quality assurancea Frequency Test Tolerance Weekly Annually Safety interlock checks Contrast resolution with captured images Safety interlock and motion checks Readout checks (e.g., source-to-detector distance) Contrast resolution Spatial resolution Review of procedures and inservice with therapists All must bc operational Meet commissioned data All must be operational 5mm 9ee Ref. [7] for recommended commissioning. Meet commissioned data Meet commissioned data All operators must fully understand operation and procedures flat-field image is taken with homogeneous phantom material placed at isocenter to provide some photon scatter and approximate the presence of a patient. Table 3 lists the QA tests for portal imaging Multileaf collimation The MLC has two carriages that each control 26 leaves that project to 1.O cm width at isocenter. The leaf positions are determined by workstation files that are delivered over a network (or floppy disk as backup) from our three-dimensional treatment planning system or a digitizer-computer system supplied by the manufacturer (Shaper, Varian Assoc., Palo Alto, CA). The desired leaf positions are then sent to the MLC controlling computer. The carriage and leaf positions are set and confirmed by digital encoders. On the MLC system if the encoders find deviations of measured and expected leaf positions greater than the designated criterion (.1 mm), an error state will not allow treatment. The leaf positions are calibrated during commissioning. Because the MLC system is an add-on device, due to the added weight leaf position accuracy and isocenter must by checked at various gantry angles by performing visual and film checks of isocenter. Leaf position is determined by the user (light edge, 5% radiation edge, etc.). Because of the curved end leaf nature on the single focused leaves, the light field edge will not correspond with the 5% radiation edge at most leaf positions and will differ from the 5% edge by as much as.5 mm. This maximum difference is observed at the position of. mm. The accuracy of position vs. light field is specified by the manufacturer to be 1.O mm over the leaf range for each leaf. As many methods and steps are in- volved with the shaping process (three-dimensional treatment planning system, Shaper, network, controller, workstation, etc.), they must also be integrated into the QA process, The commissioning of an MLC system must check position accuracy and consistency over the entire leaf span range for the clinical range of collimator and gantry angles. Decisions on acceptance criteria must be agreed on at the time of purchase. For add-on MLC units, it is imperative that the accuracy of jaw position and isocenter location are maintained despite the added weight of the MLC. The commissioning process was described by Klein et al The range of clinical uses may determine the number and frequency of tests. For example, if networking is used, checks of the stability and accuracy of data transfer must be performed, If any form of dynamic leaf motion is used, QA must also confirm parameters such as leaf speed. Testing of interlocks is mandatory. For example, routine testing must ensure that the trailing edge of leaves cannot be left unshielded by collimating jaws. Other interlocks include leaf retraction during electron therapy, portal film mode, and exiting the MLC controlling software. An important aspect of a QA program for hightechnology devices is routine inset-vices to the therapists. By performing only an introductory inservice, one is inhibited from discovering misunderstandings about the system s use. Feedback from the therapists is essential in evaluating the need for changes in written procedures and other problems. Our clinic creates MLC leaf position files by either digitization methods or by a direct generation of leaf positions by our 3D treatment planning system (3DRTP). The files are eventually transferred over a network system to the MLC controller and workstation at the treatment machine. Prior to use, each field is illuminated onto the original simulation film or digitally reconstructed radiograph (DRR). A match of light field and original shaped field must be maintained to within 2 mm for all boundaries. Once the fields have been approved, the chosen jaw settings and field name must also be checked. Inappropriate jaw settings could block a portion of the desired field, or generous settings could leave a trailing leaf region unblocked. Once the fields and relevant information are checked, the MLC file is made available for treatment. As it is impossible to check each leaf at a variety of positions over the range of gantry angles, a smaller subset must be used. On a monthly basis we check the range of leaf positions by forming diamond-shaped and X-shaped fields. We maintain these designated QA patterns for testing. These fields are configured for leaf positions stepped at l-cm intervals to allow an efficient check of light field and radiation field with a graph paper grid. An example of an X-field film superimposed over l-cm gridded graph paper is seen in Fig. 5.

7 E.E. Klein et al. /Radiotherapy and Oncology 38 (19%) 51-6 Table 4 Multi-leaf collimation quality assurancea Frequency Test Tolerance Fig. 5. MLC X-field film superimposed over I-cm gridded graph paper. Films are exposed with these fields at the cardinal gantry angles and are later scanned during monthly QA checks. A l-mm correspondence between light field (our definition of field edge) and planned position (ends and sides) is required. Treatment interlocks are also checked as part of the monthly MLC QA (Fig. 6). These interlocks were designed to prevent treatment with electrons, irradiation behind the trailing edge of leaves, irradiation for fields larger than 26 x 4 cm2 (or 4 x 26 cm2), treatment without the jaws shielding the carriage, and treatment when the requested leaf spread on a particular carriage exceeds 14.5 cm (farthest extended position vs. farthest retracted position). Retraction of leaves is confirmed when an electron applicator is inserted, exiting from the workstation software (stand alone or park mode), and entering port film mode. Transfer of tiles from one MLC machine to an another is checked on a monthly basis. Corresponding MLC fields are scanned by a computerized water scanning system with waterproof ion chambers (Dynascan, CMS, Inc., St. Louis, MO) during the annual QA check. We perform the tests listed in Patient specific Monthly Annually Check of MLC-generated light fields vs. sim film (or DRR) before each file is used Double check of MLC field settings by therapists for each fraction On-line imaging verification for most patients on each fraction Port film approval before second fraction Setting vs. light field vs. radiation field for two designated patterns Check of active patient files Testing of network system Check of interlocks Setting vs. light vs. radiation field for set patterns over range of gantry and collimator angles Water scan of set patterns Film scans to evaluate inter-leaf leakage and abutted leaf transmission Review of procedures and inservice with therapists 2mJn Expected field Physician discretion Physician discretion 1mrl.l Removal of inactive files Expected fields over network All must be operational lmrll 5% radiation edge within 1 mm Interleaf leakage c 3%, abutted leakage c 25% All operators must fully understand operation and procedures See Ref. [S] for commissioning of MLC; this table is reproduced from Ref. [S]. Table 4 for MLC. These tests can also be performed if there is any change to the MLC hardware or software. 3.Results We have varying quality assurance histories for these devices, as each came into clinical use at different times Asymmetric jaws The asymmetric jaw position accuracy has held over a 3-year period. So far the accuracy of jaw position does not depend on gantry angle, since the addition of the MLC system. This stability of position accuracy has given us confidence to use asymmetric jaws to match clinical fields [3] without the need for scheduled

8 58 E.E. Klein et al. /Radiotherapy and Oncology 38 (19%) 51-6 Month, Year: Multi-Leaf Collimation Monthly Quality Assurance Under patient file: mlcqa retrieve diamond field and go to shape. - a. With 4x4 cm2 tield set, check for carriage unshielded error and open area not under jaw error Place electron applicator in collimator assembly h. The leaves will retract (remove applicator) Reshape diamond pattern Gotoportfilm c The leaves will retract (exit prot film mode) Reshape diamond pattern Switch mode to Park d The leaves will retract Cl Cl Set 26 x26 cm2 Field Under patient tile: mlc.qa retrieve diamond field and go to shape. With graph paper, ensure the leafends project at l.ocm intervals to within lmm at 1COcm Film field under 1Scm depth at 1OOcm SFD, etching leaf ends Ensure radiation field is congruent to lmm Select inverse field and go to shape. With graph paper, ensure the leaf ends project at l.ocm intervals to within lmm at 1CKkm Film field under 1.5cm depth at 1OOcm SFD, etching leaf ends Ensure radiation field is congruent to lmm Rotate gantry to 9 and select diamond field and go to shape. With graph paper, ensure the leafends project at l.ocm intervals to within lrnm at 1OOcm Rotate gantry to 27 and select inverse field and go to shape. With graph paper, ensure the leaf ends project at l.ocm intervals to within lmm at 1OOcm QA pet-found by : Date of QA: Reviewed by: Cl Fig. 6. Multileaf collimation monthly quality assurance check form. feathering of the match line. We have demonstrated this accuracy by taking composite port films, in which adjacent fields are exposed on the same port film. When a narrow region of underdose was observed (Fig. 3), the jaw readout potentiometers were adjusted. The OAFS that have been measured periodically have held to within.5%. We have had no problem with back scattering into the monitor chamber Dynamic Wedge The dynamic wedge SITS that have been tested (plotted, profiles measured, wedge factors) periodically have not been corrupted and have remained constant over a 2-year period. The diode results have shown only one deviation greater than 3%. On this occasion, the field size was changed after the first fraction (following a physician-dictated portal change) by the therapist without alerting the physics staff for recalculation. The measured diode difference of 5% was exclusively composed of the wedge factor difference. A record-andverify system with limited edit privileges would have caught this error. The measured wedge factors have not changed in 2 years Portal vision The EPID has been operating reliably in the 1.5 years since clinical implementation. The weekly QA procedures have identified periods when the EPID was not

9 E.E. Klein et al. /Radiotherapy and Oncology 38 (19%) providing optimal images. These were due to one of two sources: (1) the accelerator was not producing a stable dose rate, yielding dose bars in the image, or (2) the sensitivity of the EPID was changing because of a drift in the high-voltage power supply that placed single pixel-wide stripes on the image. The first symptom was treated by tuning the accelerator and the second by recalibrating the EPID. The safety tests have thus far revealed no failures. Unfortunately, the current EPID cannot be used with DW, due to the EPID s calibration requiring fixed pulse lengths and constant dose rate which is not possible with DW Multileaf collimation With a year s history of QA for MLC, we have seen no major deviation from the original calibration over the range of gantry angles, leaf positions, etc. All interlocks have functioned correctly during clinical operation and monthly testing. The patient-specific QA has not shown any systematic problems with leaf contiguration or file transfer. A few human errors were caught, such as field names being swapped (for opposed laterals) or a misinterpretation of the field area being treated as opposed to blocked. One error occurred when the same shaped field was used for consecutive portals. An R&V system would have stopped this error. The monthly and annual quality assurance checks have shown the MLC to perform consistently in terms of position accuracy. We are paying close attention to the tests performed at lateral gantry angles to watch for any shifting of the device. We are monitoring interleaf leakage because of the repeated leaf interdigitation. The EPID has also been used as a verification device for the MLC. This is necessary as our machine is not equipped with an R&V system and the therapists are not entering the treatment room between MLC portals for treatments requiring a large number of daily portals. One example is the treatment of earlystage prostate cancer using six or seven conformal beams. All other patients have the light field illuminated for each port, each day. As there is no R&V system, the therapists are instructed to cross-check each other before treatment is initiated. One therapist will set the main console parameters and then check the identification of the MLC field selected. The second therapist will perform the remaining functions. The average time spent carrying out an annual calibration has increased by approximately 5% for the dual energy accelerator with the majority of the additional time dedicated to MLC. The monthly MLC check takes approximately 3 min. The weekly on-line imaging checks take approximately 2 min. The patient specific MLC check of the shaped field illuminated onto the simulation film (or DRR) takes an individual approximately 6 s per field. Any QA program takes a team effort. Our clinic s QA program is spearheaded by the physics staff but the therapists and maintenance engineers are as important. The therapists are the end users and are the personnel mainly responsible for proper operation, reporting of problems, and feedback to the engineers and physicists. In turn, it is the physicists responsibility to ensure that the therapists are comfortable with the operation and procedures related to these devices. The engineers in our clinic are not only responsible for troubleshooting problems but also for performing some of the QA procedures. The engineers should also be the liaisons with the vendors service engineers. They must also communicate all repairs or changes to the physicists and therapists. 4. Discussion and conclusions Some of the frequency and tolerances described for the QA tests are governed by TG-4 recommendations. In some cases we have called for higher frequency and tighter tolerances. The TG-4 recommendations call for jaw positional accuracy of 2 mm. We require an accuracy of 1 mm due to our field matching techniques. For dynamic wedge, we maintain the required 2% tolerance for wedge factors measured annually, however by taking an initial diode measurement for each DW field, we ensure the wedge factor is correct for the given STT. Task Group 4 recommends that field size indicators be checked monthly with a tolerance of 2 mm. We check each MLC light field (vs. indicated position) for each MLC field before clinical use. This is necessary because of the network and shaping processes involved with MLC. The introduction of these new devices into the clinic must be met with strong QA programs. The programs are even more critical if there is not an R&V system available that covers all of the dose-delivery technologies. But even the strongest QA program cannot replace an R&V system that is fully equipped to handle the new technologies. An R&V system would alleviate some of the patient specific checks, such as the field illumination following a DW treatment, or the console cross-check performed for MLC fields. The R&V system would also require a comprehensive QA program involving conventional and specialized device checks. The R&V system would have to be provided programmed according to the same tolerances described in this manuscript. Portal imaging is a supplementary QA device, but the functionality of the EPID also requires a rigorous QA program. The most important QA tool is continuing education. The process cannot stop with initial training. Continual teaching, updating of procedures, and observation can avoid persistent misconceptions of the device s operation. The one question that has continued to arise in our

10 6 E. E. Klein et al. /Radiotherapy and Oncology 38 (19%) 51-6 clinic is who should take responsibility for which QA tasks. The physicist cannot be removed totally from the process between annual checks. At the same time, many daily, weekly and monthly processes can be carried out by personnel, such as maintenance engineers, dosimetrists or therapists, if the proper training is given. Part of the learning curve is to decide the extent and frequency of the QA programs. Without question, an R&V system should be a standard component of an accelerator that possesses these new technologies. References [l] AAPM. American Association of Physicists in Medicine, Proceedings of a Symposium on Quality Assurance of Radiotherapy Equipment. AAPM Symposium No. 3. American Institute of Physics, New York, NY, [2] AAPM. American Association of Physicists in Medicine, Comprehensive QA for Radiation Oncology: Report of AAPM Radiation Therapy Committee Task Group 4. Med. Phys. 21: , [3] Klein, E.E., Taylor, M., Michaletz-Lorenz, M., Zoeller, D. and UmBeet, W. A mono isccentric technique for breast and region- al nodal therapy using dual asymmetric jaws. Int. J. Radiat. Oncol. Biol. Phys. 28: , [4] Klein, E.E., Low, D.A. and Purdy, J.A. Dosimetry for clinical implementation of dynamic wedge. Int. J. Radiat. Oncol. Biol. Phys. 31: , [5) Klein, E.E., Harms, W.B., Low, D.A., Willcut, V. and Purdy, J.A. Clinical implementation of multi-leaf collimation: dosimetry, networking, and quality assurance, Int. J. Radiat. Biol. Phys., 33: [6] Kubo, H. Telescopic measurements of backscattered radiation from secondary collimator jaws to a beam monitor chamber using a pair of slits. Med. Phys. 16: , [7] Low, DA., Klein, D.A., Maag, D. and Purdy, J.A. Commissioning and periodic quality assurance of an electronic portal imaging device. Int. J. Radiat. Oncol. Biol. Phys., 34: [8] Slessinger, E.D., Gerber, R.L., Harms, W.B., Klein, E.E. and Purdy, J.A. Independent collimator dosimetry for a dual photon energy linear accelerator. Int. J. Radiat. Oncol. Biol. Phys. 27: , [9] Starkschall, G. and Horton, J. eds. Quality Assurance in Radiotherapy Physics. Proceedings of an American College of Medical Physics. Madison, WI: Medical Physics Publishing; [lo] Thompson, A.V., Lam, K.L., Balter, J.M., McShan, D.L., Mattel, M.K., Weaver, T.A., Fraass and Ten Haken, R.K. Mechanical and dosimetric quality control for computer controlled radiotherapy treatment. Units. Med. Phys. 22: , 1995.

11 本文献由 学霸图书馆 - 文献云下载 收集自网络, 仅供学习交流使用 学霸图书馆 ( 是一个 整合众多图书馆数据库资源, 提供一站式文献检索和下载服务 的 24 小时在线不限 IP 图书馆 图书馆致力于便利 促进学习与科研, 提供最强文献下载服务 图书馆导航 : 图书馆首页文献云下载图书馆入口外文数据库大全疑难文献辅助工具