Home Search Collections Journals About Contact us My IOPscience An accurate calibration method of the multileaf collimator valid for conformal and intensity modulated radiation treatments This article has been downloaded from IOPscience. Please scroll down to see the full text article. 2004 Phys. Med. Biol. 49 2631 (http://iopscience.iop.org/0031-9155/49/12/011) View the table of contents for this issue, or go to the journal homepage for more Download details: IP Address: 152.92.171.92 The article was downloaded on 15/10/2010 at 15:39 Please note that terms and conditions apply.
INSTITUTE OF PHYSICS PUBLISHING Phys. Med. Biol. 49 (2004) 2631 2643 PHYSICS IN MEDICINE AND BIOLOGY PII: S0031-9155(04)71864-3 An accurate calibration method of the multileaf collimator valid for conformal and intensity modulated radiation treatments Maria Sastre-Padro, Uulke A van der Heide and Hans Welleweerd Department of Radiotherapy, University Medical Center Utrecht, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands E-mail: u.a.vanderheide@radth.med.uu.nl Received 11 November 2003 Published 2 June 2004 Online at stacks.iop.org/pmb/49/2631 doi:10.1088/0031-9155/49/12/011 Abstract Because for IMRT treatments the required accuracy on leaf positioning is high, conventional calibration methods may not be appropriate. The aim of this study was to develop the tools for an accurate MLC calibration valid for conventional and IMRT treatments and to investigate the stability of the MLC. A strip test consisting of nine adjacent segments 2 cm wide, separated by 1 mm and exposed on Kodak X-Omat V films at D max depth, was used for detecting leaf-positioning errors. Dose profiles along the leaf-axis were taken for each leaf-pair. We measured the dose variation on each abutment to quantify the relative positioning error (RPE) and the absolute position of the abutment to quantify the absolute positioning error (APE). The accuracy of determining the APE and RPE was 0.15 and 0.04 mm, respectively. Using the RPE and the APE the MLC calibration parameters were calculated in order to obtain a flat profile on the abutment at the correct position. A conventionally calibrated Elekta MLC was re-calibrated using the strip test. The stability of the MLC and leafpositioning reproducibility was investigated exposing films with 25 adjacent segments 1 cm wide during three months and measuring the standard deviation of the RPE values. A maximum shift over the three months of 0.27 mm was observed and the standard deviation of the RPE values was 0.11 mm. 1. Introduction Increasing numbers of accelerators are nowadays equipped with a multileaf collimator (MLC) and used for conventional treatments, replacing metal alloy blocks. Several papers compare the dosimetry of the MLCs and blocks used in conventional 3D-radiotherapy (Boyer et al 1992, Helyer and Heisig 1995, LoSasso and Kutcher 1995, Palta et al 1996, Meyer et al 1999, Chang et al 2000, Cheng et al 2001). In general, the dosimetry of the radiation fields is 0031-9155/04/122631+13$30.00 2004 IOP Publishing Ltd Printed in the UK 2631
2632 M Sastre-Padro et al similar whether shaped by the MLC or by the metal alloy blocks. When using the MLC for intensity modulated treatments (IMRT) the implications of transmission through the MLC must be considered thoroughly (Galvin et al 1993, Chui et al 1996, Wang et al 1996, Boyer and Li 1997, LoSasso et al 1998, Chen et al 2000, Kung and Chen 2000, Sharpe et al 2000, Deng et al 2001, Graves et al 2001). In IMRT treatments the transmissions contribute to the dose throughout the target and not just on the field boundary. Leaf ends not only delimit the field, as in conventional 3D treatments, but also are projected into the target affecting the dose inside the target volume. The effect of random leaf-positioning inaccuracies will be blurred over several treatment sessions and could compromise steep dose-gradient areas between the target and organ at risks. The effect of systematic leaf-positioning inaccuracies will be added during the treatment of a patient and can generate large deviations from the expected dose inside the target. Many authors have addressed leaf-positioning accuracy and MLC-performance (Jordan and Williams 1994, Chui et al 1996, Hounsell and Jordan 1997, Budgell et al 2000, Low et al 2001, Samant et al 2002, Vieira et al 2002, Bayouth et al 2003, Yang and Xing 2003). One of the most important systematic errors that, probably, will not be noticed during treatment, is a wrong calibration of the leaves. A calibration method that is not accurate enough or a non-appropriate leaf-positioning definition can result in an incorrect leaf positioning. Using a large number of (small) segments can cause a relevant dose error. Improvement of the conventional calibration method will allow tightening of the tolerances on the leaf positions. (Tolerances on leaf position of the Elekta MLC as specified by the manufacturer are 1 mm or 1%, whichever is greatest.) The quality of a calibration procedure relies on the accuracy with which leaf-positioning errors can be detected. Methods that are most commonly used are direct measurements of the light field and measurements with an ionization chamber in a water tank. The first method is simple but requires corrections because of the differences between light field and radiation field produced by the rounded leaf ends (Kung and Chen 2000). The use of the water tank as calibration method implies a leaf positioning defined at certain dose percentage. For the jaws the 50% dose is used and for the leaves also it is a common choice. The central leaves are calibrated with the water tank and the others are visually aligned to the central leaves using the light field. Both methods are quite time consuming. Moreover, depending on the MLC position in the accelerator head, the blurred light field edge may hamper an accurate alignment of the leaves. In this study we developed the tools for an accurate calibration of the MLC. We used a strip test for detecting leaf-positioning errors and performed a quantitative analysis of the test that can be used directly for calibrating the MLC. A strip test consists of a series of narrow adjacent segments shaped by the leaves, irradiated on a film. Because on the abutment regions (area in between two adjacent segments) the dose delivered is very sensitive to positioning errors, the use of adjacent segments is a good approach to measure leaf-positioning errors and has been widely used before (Chui et al 1996, Chui and LoSasso 2000, Low et al 2001). We intend to obtain a leaf positioning that produces a flat profile on the abutment between two segments. The alignment obtained using this definition is compared with the 50% dose definition used conventionally. Because the accelerators equipped with MLC are generally used for both conventional and IMRT treatments, a calibration procedure that suits both modalities should be applied. We investigated the sensitivity of the strip test by creating intentional leaf-pair misalignments. The results of the strip-test analysis were used to recalibrate an Elekta MLC. Finally we investigated the stability of the MLC and quantified the leaf-positioning reproducibility.
Accurate MLC calibration for IMRT 2633 2. Methods and materials 2.1. Quantitative analysis of a strip test A common method of measuring the leaf-positioning accuracy is the irradiation of a number of adjacent segments on a film. A strip test consists of a series of narrow adjacent segments shaped by the leaves, irradiated on a film. Because on the abutment regions (area in between two adjacent segments) the dose delivered is very sensitive to positioning errors, the use of a strip test is a good approach to measure leaf-positioning errors. In most of the cases leaf-pair positioning errors are visually detected (Chui et al 1996, Chui and LoSasso 2000). In this study we developed a method for quantifying the leaf-pair positioning accuracy and derive the calibration parameters of the MLC. The positioning accuracy of the leaf-pair j creating the abutment i is quantified by a parameter that we call relative positioning error (RPE): RPE i,j = d i,j K (1) where d i,j represents the relative dose change on the abutment and K is the calibration factor that converts this relative dose change into millimetres (mm) of positioning error. d i,j is defined as the relative dose change on the abutment i for the leaf-pair j: ( ) D i,j d i,j = 1 (2) D i,j with D i,j being the extreme dose at the abutment and D i,j the average dose at the centre of the two adjacent segments. The RPE i,j sign is positive when an underdose (D i,j D i,j ) occurs in the abutment and negative when an overdose (D i,j D i,j ) occurs. The relative positioning error of the leaf-pair j is calculated as the average of all abutments from that leaf-pair: ( ) n 1 RPE j = RPE i,j. (3) n i=1 The RPE quantifies the relative positioning accuracy of the leaf-pair when creating the abutment. But, when both leaves of the leaf-pair are shifted in the same direction, the same dose will be delivered on the abutment and the RPE j will not change. To detect these shifts we introduced a second parameter that measures the absolute positioning error. We defined µ i,j (measured position of the abutment i created by the leaf-pair j) as the position of the extreme dose D i,j. For the leaf-pair j on the abutment i we defined the absolute-positioning error as APE i,j = ε i,j µ i,j (4) where ε i,j is the expected abutment position on the film and µ i,j the measured one. The absolute positioning error for the leaf-pair j is calculated as the average of all abutments: ( ) n 1 APE j = APE i,j. (5) n i=1 These parameters can be used for calibrating an MLC. The specific procedure will depend on the MLC type that is used, and, in this study, it is applied to an Elekta MLC i.
2634 M Sastre-Padro et al Back-up collimator Abutment Back-up collimator Leaf Figure 1. A strip-test design for calibration purposes. Nine adjacent segments 2 cm wide with a gap of nominally 1 mm and two extra segments with four squares to determine the isocentre. 2.2. A strip-test design for calibration purposes In the previous section we have formulated a quantitative analysis of a strip test that can be used directly for calibrating the MLC. For calibration purposes, we designed a strip test consisting of nine adjacent segments 2 40 cm 2 with an intentional gap of nominally 1 mm in order to have a measurable dose change on the abutment (see figure 1). The measured position of the abutment µ i,j, used to calculate the absolute positioning error APE i,j (equation (4)), is very sensitive to the definition of the isocentre position in the leaf motion direction. Due to the limited accuracy of the lasers (±1.0 mm) used for positioning the film on the treatment table, we determined the isocentre exposing one specific segment two times with a 180 collimator angle difference (figure 1). This segment consists of four squares 3 2.5 cm 2 exposed close to the border of the film. The back-up collimator delimits the segment in the outer side of the film and the leaves in the inner side. For each square we measured the 50% dose point of the outer boundary in the direction of the leaf motion. We calculated the central position for each two opposite diagonal squares (first-right square with last-left square, second-right square with the third-left and so on). We defined the isocentre position as the average of these four central values. To check the reproducibility of this method on finding the isocentre, we repeated this procedure on the same film four times and found a standard deviation of the values of 0.06 mm. Because the film may be slightly rotated during irradiation or scanning, we determined the rotation based on the position of the squares irradiated with the same collimator angle as the strips and corrected for it.
Accurate MLC calibration for IMRT 2635 80 70 D ij = dose calculated as the average Dose (cgy) 60 50 40 30 20 10 D i,j = minimum dose µ i,j = measured position of the abutment 0-100 -75-50 -25 0 25 50 75 100 Position referred to isocentre (mm) Figure 2. Analysis of the strip test. Dose profiles were taken for each leaf-pair. This figure shows the profile of a central leaf. The relative dose variations on the abutments were used to determine RPE i,j, and the measured position of the abutments was used to determine APE i,j. After scanning and converting the film to dose, we took a dose profile for each leaf-pair and measured D i,j, D i,j and µ i,j to calculate RPE i,j and APE i,j on each abutment. The average dose of the two adjacent segments, D i,j, was calculated as the average value over a range of 2 to +2 mm around the centre of the segments (see figure 2). 2.3. Film dosimetry All measurements were performed using Kodak X-Omat V films of 33 41 cm 2. The films were placed at the isocentre distance (100 cm) and exposed with a 10 MV photon-beam at a depth of D max. A polystyrene plate of 2 cm was placed on top of the film as a build-up. To find the optical density to dose conversion, we exposed two 60 wedge 20 20 fields with a difference of 180 collimator rotation with a 10 MV photon-beam. The average values of the optical density of these two films were assigned to the dose measurements in a water tank and fitted to a fifth-order polynomial function. For a complete coverage of the film dose range, two sets of two films, one with 100 MU and other with 500, were exposed so that we obtained overlapping curves. The dose coverage of this calibration was from 0 to 124 cgy. With this method we also determined the significance of the beam hardening effect. The end of the low MU curve, which corresponds to the thin area of the wedge, overlaps the beginning of the high MU curve, which corresponds to the thick part of the wedge. The discrepancy of the two curves on the overlapped area indicates the significance of the beam hardening effect. Our measurements showed an excellent agreement on the overlapped area suggesting no significant influence of beam hardening on the film calibration. The films were scanned on a Lumiscan scanner with a pixel size of 0.4 mm. Scan corrections in the scan direction were used to avoid geometrical distortions associated with the scanning process. To determine the geometric accuracy of the scan corrections we scanned a 35 cm ruler and measured the distances between the central cm mark of the ruler and the others. These measurements were performed with and without scan corrections. In figure 3 the distance differences between the measured distance and the expected one are plotted for both cases. The scan corrections performed correct for deviations up to 2.2 mm with an accuracy below 0.1 mm.
2636 M Sastre-Padro et al 2.5 SCAN CORRECTIONS ON THE LUMISCAN x measured - x expected (mm) 2 1.5 1 0.5 0-0.5 No Corrected Corrected -1 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 data point (a.u.) Figure 3. Film dosimetry. Scan corrections to avoid geometrical distortions of the Lumiscan scanner. A 35 cm ruler placed on the scan direction was scanned with 0.4 mm pixel size. We measured the distance between the central cm mark and the others. The graph shows the difference between the measured distance and the expected one with and without corrections. 2.4. Machine performance In this study, we used a different strip-test design to measure the reproducibility of the leaf positioning of the Elekta MLC. We irradiated films with 25 adjacent segments 1 40 cm 2. From the abutments, we measured the relative positioning error between the two leaves of the leaf-pair, for every leaf-pair and every abutment. We did this measurement during 20 consecutive days and after that once every week for a period of three months. The variation of the relative positioning error (RPE) with time relates to the frequency of quality assurance required, guaranteeing an optimal performance of the MLC. To detect the influence of leaf motion direction and travel distance on the leaf-positioning accuracy, we implemented two versions of the strip test: one with a sliding series of adjacent segments (leaves move from one segment to the next adjacent one), and another with a pre-determined but randomly placed series of segments. 3. Results 3.1. Determination of the RPE calibration factor, K In equation (1) the calibration factor K converts the relative dose change in the abutment d i,j into millimetres (mm) of relative positioning error RPE and has units of mm. The K value was determined experimentally by creating five strip tests with different leafbank misalignments. Each leaf bank was shifted 0.25, 0.5, 0.75, 1 and 1.25 mm in opposite directions. These misalignments resulted in different dose variations on the abutments, modifying the RPE i,j in 0.5, 1, 1.5, 2 and 2.5 mm, respectively. Plotting the introduced nominal RPE versus the measured relative dose change on the abutment, d i,j, we obtained a curve with a slope of 9.85 mm (figure 4). The point with d i,j = 0 corresponds by definition to an RPE = 0. The misalignments that are used to find the K value suffer from the original inaccuracies in leaf alignment. For an Elekta MLC this means that both leaf offset and gain may be inaccurate. A dose variation in the profile is created by two opposing leaves (from two adjacent strip segments) at nearly the same position. As a result an incorrect gain may shift the
Accurate MLC calibration for IMRT 2637 3.0 K experimentally determined RPE (mm) 2.5 2.0 1.5 1.0 0.5 0.0-0.5 RPE = 9.85 * d 0 0.05 0.1 0.15 0.2 0.25 0.3 d Figure 4. Experimental determination of the RPE calibration factor, K.The K value was calculated as the slope of the curve RPE versus d in the range of [0, 2] mm RPE. entire abutment, but the effect on the relative dose change is negligible. An incorrect offset on the leaves will have an effect on the total dose change. However, this effect will be constant for the entire series of misalignments used to determine the K factor. Thus, the slope of the curve (figure 4) is not affected. 3.2. The strip-test accuracy When performing a strip test several factors such as set-up errors, leaf-repositioning errors, film scanning errors, quantitative-analysis errors, etc, limit our accuracy in detecting leafpositioning misalignments. To determine the accuracy of the film scanning and the quantitative analysis, we scanned and quantified the same film four times. In this case, variations in set-up and leaf repositioning were not affecting our results. We calculated the differences of each individual value with the average calculated from the four scans, RPE i,j RPE i,j and APE i,j APE i,j. The reproducibility of the measurement on finding the same RPE and APE, measured as the standard deviation of these values, was 0.04 mm and 0.15 mm, respectively. To investigate the accuracy of the strip test finding the RPE i,j we used the set of measurements presented in the previous section to measure the K value. We intentionally introduced leaf-pair misalignments that created RPEs of nominally 0.5, 1.0, 1.5, 2.0 and 2.5 mm. For each film we calculated the differences RPE i,j RPE expected. The standard deviation of these 1600 numbers (40 leaves 8 abutments 5 films) reports about the accuracy of the test on finding the misalignment introduced. The overall standard deviation was 0.20 mm, in figure 5 a histogram with all the RPE i,j RPE expected is shown. To investigate the accuracy of the strip test finding the APE i,j, we shifted each leaf bank 2 mm in the same direction and compared the APE i,j withanoshiftedsituationusedasa reference APE reference. The standard deviation of the values APE i,j APE reference was 0.23 mm and the average value was 2.03 mm (expected 2.0 mm that was the shift introduced). 3.3. The strip test to calibrate the Elekta MLC As mentioned in previous sections, the strip test can be used to find the absolute and relative positioning errors (equations (1) to(5)). The next step is to find the calibration parameters that correct the measured leaf-positioning errors for a specific MLC. An Elekta SL20 equipped with an integrated multileaf collimator was calibrated using the quantitative analysis and the
2638 M Sastre-Padro et al Frequency 350 300 250 200 RPE histogram 150 100 50 0-1 -0.8-0.6-0.4-0.2 0 0.2 0.4 0.6 0.8 1 RPE i,j - RPE expected Figure 5. The strip-test accuracy. Frequency of the RPE i,j RPE expected values measured when misalignments in the range of 0.5 to 2.5 mm in steps of 0.5 mm were introduced in the leaves. strip test described in the previous sections. Jordan and Williams give an extended description of the Elekta MLC i (Jordan and Williams 1994). The calibration parameters of the Elekta MLC i are one gain for the whole MLC, general offsets for each leaf bank and individual offsets for each leaf. The gain is defined as the slope of the linear relationship between the leaf position set in the MLC control (desired) and the actual leaf position (measured). With a wrong gain factor the position of the leaf at the central axes could be correct, but as its position moves away from the centre of the field, the leaf-positioning error will increment linearly. To calculate the gain we used a similar method to the one described by Budgell et al (2000). They measured the leaf positioning on both sides of the central axis, subtracted these measured positions and divided them by the expected size. In this study we calculated the gain for each leaf as the difference of the measured position between the first and the last abutment divided by the difference of the expected ones. The gain for the jth leaf-pair in a series of eight abutments is g j = µ 1,j µ 8,j. (6) ε 1,j ε 8,j Because only one gain is defined for the whole MLC, its value is averaged over the 40 leaf-pairs 40 j=1 G = g j. (7) 40 The offsets are defined as a constant leaf-shift along the field and can be general offsets (resulting in an identical shift for all the leaves of a leaf bank) or individual offsets (shifts for each leaf separately). Combining the relative positioning error, RPE j, and the absolute positioning error, APE j, we calculated the individual offsets for each individual leaf: l o j = RPE j 2 +APE j (8) r o j = RPE j +APE j. (9) 2 The first term represents the misplacement of the leaves in opposite directions that causes the dose variation on the abutment. The second term represents the misplacement of the leaves in the same direction that shifts the abutment with respect to the expected position.
Accurate MLC calibration for IMRT 2639 0.4 APE values after strip-test calibration APE (mm) 0.2 0.0-0.2-0.4 0.4 Leaf-pair RPE values after strip-test calibration RPE (mm) 0.2 0.0-0.2-0.4 Leaf-pair Figure 6. The strip test to calibrate the Elekta MLC. This figure shows the APE j and RPE j for each leaf-pair after the strip-test calibration. The standard deviations of the values were, in average, 0.23 and 0.18 for the APE j and the RPE j, respectively. The general offsets are calculated as an average of the individual offsets: 40 j=1 L O = l o j 40 40 j=1 R O = r o j. (11) 40 To calibrate the MLC we performed a strip test and derived the gain factor (equations (6) and (7)), the individual offsets (equations (8) and (9)) and the general offsets (equations (10) and (11)). A gain error shifts the position of the leaves, in theory, by a linear relationship. Therefore, using one strip test we could calculate the offsets required after a gain adjustment. In practice, after a gain adjustment, we preferred to use a new strip test for the offset calculations. Then, we corrected for the general offsets (corrections of the leaf-banks positioning) and, finally, carried out individual corrections for each leaf calculated as the difference between the general offset and the individual offset. In the calibration of the Elekta MLC specific units are used to change the offset values. Corrections below one unit (6 units mm 1 ) are not possible. This value is in the same range of the strip-test accuracy and limits the accuracy achievable when performing a strip-test calibration. We calibrated the MLC based on the strip test and the new leaf-positioning definition. After the calibration, we performed a new strip test. In figure 6 the APE j and RPE j values after the calibration are shown. The standard deviation of the values was 0.23 mm for the APE and 0.18 mm for the RPE. (10)
2640 M Sastre-Padro et al 120% Effect of the 50% dose calibration in abutting segments 100% profile over a central leaf Dose (cgy) 80% 60% 40% 50% profile shifted to match at 50% addition 20% 0% 1 11 21 31 41 51 61 71 81 91 101 111 121 131 141 151 Position (a.u.) Figure 7. Simulation of the abutment between two adjacent segments when leaf positioning is defined at 50% dose. While individual offsets are used for each leaf, one average gain value is used for the entire MLC. The standard deviation of the individual gains (equation (6)) was found to be around 0.003 (3 mm on 1 m). Because the MLC is at the same time used for conventional and IMRT therapy, it is important to know the effect of the new leaf-positioning definition in conventional treatments. A conventional calibration for the Elekta MLC implies a leaf-positioning definition based on the 50% dose that differs from the one used in the strip-test calibration, based on a flat profile on the abutment of 2 cm wide adjacent segments. We simulated the conventional calibration adding the penumbras of the right and left leaves at the 50% dose. An RPE of 0.28 mm was found in the abutment, requiring each leaf bank to be retracted with a general offset of 0.14 mm to obtain a flat profile (see figure 7). After the strip-test calibration, we measured the radiation field of 10 10 and 30 30 cm 2 only-leaves fields. The radiation field was exposed on Kodak X-Omat films and the measured field size was obtained from several profiles at the central position. The accuracy of the measurements was estimated to be ±0.4 mm (1 pixel). The radiation fields yielded 300.3 mm and 99.7 mm, for the 30 30 and 10 10 fields, respectively. Therefore, the impact of this new leaf-positioning definition in conventional treatments is not relevant and the strip-test calibration suits both modalities, conformal and IMRT. 3.4. Machine performance During three months we exposed two films with 25 segments of 1 40 cm, one with sliding segments and the other with random segments. For this design of the strip test the K value was calculated. For each leaf-pair we calculated the standard deviation of the RPE i,j values over the three months, resulting in the standard deviation of 40 leaves 24 abutments. The leaf-pair reproducibility was calculated as the average value over the 24 abutments and is shown in figure 8. The average leaf-pair reproducibility was 0.11 mm for both sliding and random sequences. No trend was observed for the sliding neither for the random segment sequences, indicating that leaf-positioning accuracy, for the Elekta MLC i, is independent of leaf-motion direction and travel distance. The RPE drift during the three months was calculated as the difference between the average RPE j of the first four days and the average RPE j of the last four days. The drift found was not more than 0.27 mm which is about the accuracy of the method. Again, no
Accurate MLC calibration for IMRT 2641 stdev RPE (mm) stdev RPE (mm) Leaf-pair reproducibility random sequence 0.30 0.25 0.20 0.15 0.10 0.05 0.00 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 Leaf-pair Leaf-pair reproducibility sliding sequence 0.30 0.25 0.20 0.15 0.10 0.05 0.00 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 Leaf-pair Figure 8. Machine performance. Reproducibility of the leaf-pair positioning measured as the standard deviation of the RPE i,j over three-month measurements. The error bars for each leaf-pair correspond to the standard deviation over the 24 abutments. difference appeared between the sliding and random sequences. This result indicates the stability of the Elekta MLC and, referred to the frequency of the MLC controls, we can assert that leaf-positioning revisions performed once every three months will not compromise the optimal performance of the MLC. 4. Conclusion and discussion When performing IMRT it is important to calibrate the leaf positioning not only at the right spot but also accurately. Water-tank measurements have a reproducibility of 1 mm on finding the same leaf position from one day measurement to another. When performing conventional therapy this accuracy is far enough, however, for IMRT, leaf deviations of 1 mm are systematic errors that could overexpose organ at risks or underexpose a boost. The strip-test calibration has a reproducibility below 0.2 mm for the leaf positioning, and aligns the 40 leaves in a straight line that is difficult to reach when the light field is used. Also, the time employed during the strip-test calibration is considerably reduced when compared with the water-tank calibration. Related to the machine performance we measured a leaf-pair reproducibility of 0.2 mm. Jordan and Williams (1994) measured the individual leaf reproducibility for an Elekta MLC using a water tank and an electronic portal imaging device (EPID) and alternative settings of 4 and 30 cm fields. Their results showed 0.4 mm of reproducibility. Bayouth et al (2003) developed a technique to measure the absolute position of each MLC leaf based on parallel strips 1 28 cm 2 wide and having a 2 cm distance between them, where the actual position of the leaf was defined by the 50% dose position. In our method
2642 M Sastre-Padro et al the position of the leaf is based on the accuracy of the leaf-pair creating the gap and we have demonstrated that for rounded leaf-end designs the calibration at the 50% dose is not the most appropriate when performing IMRT treatments. Huq et al (2002) measured the 20 80% penumbra variations along the field for the Elekta MLC. Their results showed that when the leaves are in off-axis positions the penumbra size is bigger and gets smaller when the leaves approach the central axis and are across it. In our strip-test design we have a series of adjacent segments where the leaves cover a range of ±9 cm, therefore penumbra variations will change our K value. We have not considered this effect in our measurements. In conclusion, we have introduced a calibration method based on a strip test that can be used for any MLC vendor and used it to derive the calibration parameters of an Elekta MLC i. And we have investigated the stability and leaf-positioning reproducibility of the Elekta MLC i. Acknowledgment This research was supported by an EDRO research fellowship from the European Society of Radiotherapy and Oncology (ESTRO). References Bayouth J E, Wendt D and Morrill M 2003 MLC quality assurance techniques for IMRT applications Med. Phys. 30 743 50 Boyer A L, Ochran T G, Nyerick C E, Waldron T J and Huntzinger C J 1992 Clinical dosimetry for implementation of a multileaf collimator Med. Phys. 19 1255 61 Boyer A L and Li S 1997 Geometric analysis of light-field position of a multileaf collimator with curved ends Med. Phys. 24 757 62 Budgell G J, Mott J H, Williams P C and Brown K J 2000 Requirements for leaf position accuracy for dynamic multileaf collimation Phys. Med. Biol. 45 1211 27 Chang S X, Cullip T J and Deschesne K M 2000 Intensity modulation delivery techniques: Step and shoot MLC auto-sequence versus the use of a modulator Med. Phys. 27 948 59 Chen Y, Boyer A L and Ma C M 2000 Calculation of x-ray transmission through a multileaf collimator Med. Phys. 27 1717 25 Cheng C, Das I J and Steinberg T 2001 Role of multileaf collimator in replacing shielding blocks in radiotherapy Int. J. Cancer 96 385 95 Chui C, Spirou S and LoSasso T 1996 Testing of dynamic multileaf collimator Med. Phys. 23 635 41 Chui C S and LoSasso T 2000 Quality assurance of IMRT. Focus session: IMRT delivery theoretical concepts and clinical reality XIIIth ICCR 2000 (Heidelberg, Germany) pp 168 9 Deng J, Pawlicki T, Chen Y, Li J, Jiang S B and Ma C M 2001 The MLC tongue-and-groove effect on IMRT dose distributions Phys. Med. Biol. 46 1039 60 Galvin J M, Smith A R and Lally B 1993 Characterization of a multileaf collimator system Int. J. Radiat. Oncol. Biol. Phys. 25 181 92 Graves M N, Thompson A V, Martel M K, McShan D L and Fraass B A 2001 Calibration and quality assurance for rounded leaf-end MLC systems Med. Phys. 28 2227 33 Helyer S J and Heisig S 1995 Multileaf collimator versus conventional shielding blocks: a time and motion study of beam shaping in radiotherapy Radiother. Oncol. 37 61 4 Hounsell A R and Jordan T J 1997 Quality control aspects of the Philips multileaf collimator Radiother. Oncol. 45 225 33 Huq M S, Das I J, Steinberg T and Galvin J M 2002 A dosimetric comparison of various multileaf collimators Phys. Med. Biol. 47 N159 70 Jordan T J and Williams P C 1994 The design and performance characteristics of a multileaf collimator Phys. Med. Biol. 39 231 51 Kung J H and Chen G T Y 2000 Intensity modulated radiotherapy dose delivery error from radiation field offset inaccuracy Med. Phys. 27 1617 22
Accurate MLC calibration for IMRT 2643 LoSasso T and Kutcher G J 1995 Multileaf collimator versus alloy blocks: analysis of geometric accuracy Int. J. Radiat. Oncol. Biol. Phys. 32 499 506 LoSasso T, Chui C and Clifton Ling C 1998 Physical and dosimetric aspects of a multileaf collimator system used in the dynamic mode for implementing intensity modulated radiotherapy Med. Phys. 25 1919 27 Low D, Sohn J W, Klein E E, Markman J, Mutic S and Dempsey J F 2001 Characterization of a commercial multileaf collimator used for intensity modulated radiation therapy Med. Phys. 28 752 56 Meyer K H, Skubic S E, Baird L C, Virudachalam R and Asche D R 1999 Clinically shaped fields: relative outputs for Varian multileaf collimators and tray-mounted cerrobend blocks Med. Phys. 26 1257 9 Palta J R, Yeung D K and Frouhar V 1996 Dosimetric considerations for a multileaf collimator system Med. Phys. 23 1219 24 Samant S et al 2002 Verification of multileaf collimator leaf positions using an electronic portal imaging device Med. Phys. 29 2900 12 Sharpe M B, Miller B M and Wong J W 2000 Compensation of x-ray beam penumbra in conformal radiotherapy Med. Phys. 27 1739 45 Vieira S, Dirkx M L P, Pasma L and Heijmen B J M 2002 Fast and accurate leaf verification for dynamic multileaf collimation using an electronic portal-imaging device Med. Phys. 29 2034 40 Wang X, Spirou S, LoSasso T, Stein J, Chui C and Mohan R 1996 Dosimetric verification of intensity-modulated fields Med. Phys. 23 317 27 Yang Y and Xing L 2003 Using the volumetric effect of a finite-sized detector for routine quality assurance of multileaf collimator leaf positioning Med. Phys. 30 433 41