The British Journal of Radiology, 75 (2002), 428434

2002 The British Institute of Radiology

Abutment region dosimetry for the monoisocentric three-beam split eld technique in the head and neck region using asymmetrical collimators
1

K ABDEL-HAKIM, MD, 1T NISHIMURA, MD, 2M TAKAI, PhD, 1S SUZUKI, RT and 1 H SAKAHARA, MD

Departments of 1Radiology and 2Informatics, Hamamatsu University School of Medicine, 1-20-1 Handayama, Hamamatsu, 431-3192 Japan

Abstract. Creating non-divergent eld edges using asymmetric collimators and a single isocentre can improve matchline dosimetry owing to decreased reliance on operator skills and avoidance of couch movement. However, asymmetic jaws have an associated tolerance that can cause abutment to be misaligned. The matching area dose for monoisocentric three-beam split elds commonly used in head and neck cancer treatments using mismatched and matched collimators is the subject of this work. X-ray verication lm was exposed in a solid-water phantom, and the dose at the matching area was evaluated using mismatched and matched collimators. In the case of mismatched (consistently overlapped) collimators, digital displays of an asymmetric collimator position within the tolerance indicated in the manufacturers specications were investigated for the three-beam split eld technique. The effect of this technique on the junctional dose was also determined using matched collimators. Although the collimators showed a consistent overlap, a perfect dose distribution could be obtained at the matching area. The three-beam split eld technique yielded an 8% overdose at the matchline using matched collimators. In conclusion, an awareness of the effects of the abutting technique and digital display tolerance is necessary to achieve good junction uniformity using asymmetric collimators. Asymmetric collimators or jaws are now available on modern linear accelerators. These are collimators that allow for the independent movement of each of the four eld-dening jaws. The resultant treatment eld is centred off the rotational axis of the collimator. Many investigators have studied dosimetry for asymmetric collimators [16]. Asymmetric collimators have many clinical applications such as beam splitting, planned boosting, eld reductions, matching of divergent elds, creation of multiple asymmetric elds, arc rotation, matching line feathering and production of opposed tangential elds [7]. One of the most popular applications of asymmetric collimators is eld matching, in which two split elds are matched at the central beam axis (junction) using a single isocentre. This monoisocentric set-up is commonly used in treatments of the head and neck region (two parallel-opposed lateral upper neck elds are abutted to an anterior supraclavicular eld). The technique obviates the need for movement of the couch to abut the elds and is therefore theoretically associated with more reproducible dosimetry in the junction plane. An additional and important potential gain with this
Received 9 April 2001 and in nal form 21 November 2001, accepted 29 November 2001. 428

technique is the reduced time required for set-up and treatment [8]. A linear accelerator manufacturer currently species 1.0 mm for collimator positional accuracy as an acceptance criterion. Within this tolerance, collimators can underlap or overlap at the junction of abutted elds, resulting in a dosimetric problem. Many investigators have studied the matching area dose by creating gaps and overlaps within the specied tolerance. They concluded that collimator misalignment could produce inhomogeneities of up to 40% [9, 10] or 60% [11] above or below the prescribed dose. However, there is a lack of data showing whether this misalignment is consistent or subject to random uctuations. Therefore the day-to-day variation of dose that may occur in the plane of the junction cannot be accurately predicted. Although rigorous quality assurance is regularly performed to ensure that the mechanical and dosimetric integrity of our equipment is maintained over time, one linear accelerator showed a consistent collimator overlap, hence there was a consistent cold spot at the matching area when abutting split elds. This nding was reported previously in the work of Lee [12]. Also, at the Mayo Clinic, six linear accelerators were evaluated [11]. All six machines had consistently
The British Journal of Radiology, May 2002

Abutment using asymmetrical collimators

overlapping collimators with a resultant underdose ranging from 8% to 25%. These studies investigated the matching area dose using a double-exposure technique, i.e. two-beam split eld. The aim of the present work was to investigate and improve the matching area dose for the monoisocentric three-beam split elds commonly used in head and neck irradiation using mismatched (consistently overlapped) collimators. The effect of the three-beam split eld technique on junctional dose using matched collimators was also studied.

based on the calibration curves obtained from calibration lms.

Positional accuracy of asymmetric collimators

The accuracy and precision of Y-independent jaws at the zero position was evaluated using the light eld and double-exposure technique. Computer positioning of Y1 and Y2 at the junction of two-beam split elds was rst checked against light-eld projection onto a straight line drawn on the lm jacket. In the double exposure technique, the lm was rst irradiated with Y1 at the zero position and Y2 at 10 cm. Then, collimator positions were reversed and the lm was irradiated again. To assess reproducibility, this procedure was repeated ve times on both machines. On each occasion the jaws were subjected to multiple repetitions of opening and returning to the zero position.

Materials and methods

Two of our linear accelerators (Linac 2100C and 600C; Varian Associates, Palo Alto, CA) have the capability of independent jaw movements. The machines were tested and found to be within the specications of the manufacturer and The American Association of Physicists in Medicine (AAPM) task group 40 [13]. The upper pair of jaws, closest to the beam source, is designated as the Y-set (Y1, Y2). The orthogonal jaws, further from the source, are designated as the X-set (X1, X2). Both the Y and X collimators can over-travel the central axis by 10 cm and 2 cm, respectively, and are therefore capable of creating true nondivergence. All lm dosimetry measurements were made with Kodak X-Omat VR lm (Kodak, Rochester, NY). Following radiation delivery to the lm, the optical density was normalized to a point 3 cm away from the beam central axis in the superiorinferior direction. Optical density proles were measured using a lm scanner (Microdensitometer 2405; Abe Sekkei Co., Tokyo, Japan) with an aperture size of 0.1 mm61.0 mm and a scanning pitch of 0.1 mm and converted to doses

Study 1: assessment of the matching area dose in the three-beam split eld technique using overlapped collimators
Matching was studied using a 4 MV photon beam from the 2100C Linac. An upper set of asymmetric collimators was used in this study. The monoisocentric three-beam split elds were set as shown in Figure 1a. A packed therapy verication lm was inserted between two 306 3065 cm3 solid-water slabs. The phantom blocks were positioned upright with the lm plane at a single isocenter and at a depth of 5 cm for both the upper lateral parallel-opposed elds and the lower anterior eld. We chose the eld sizes that are most frequently used in clinical settings, namely 10 cm612 cm and 10 cm620 cm for

Figure 1. Field arrangement and lm positioning with respect to a solid-water phantom. The British Journal of Radiology, May 2002 429

K Abdel-Hakim, T Nishimura, M Takai et al

lateral and anterior elds, respectively. The lateral eld was dened by closing the inferior half of the asymmetric jaws (Y150 cm, Y2510 cm, X5 12 cm). The lm was irradiated with 25 monitor units (MU) laterally at gantry angles of 90 and 270 . Then, the superior half of the eld was closed and the inferior half was opened (Y15 10 cm, Y250 cm, X520 cm) to irradiate the anterior eld with 50 MU. Measurements were made by varying the position of the Y jaws at the matching area, as in Table 1. For the purpose of comparison, the two-beam split eld technique with a xed gantry angle of 0 was also studied. The set-up is shown in Figure 1b. The superior eld was dened by closing the inferior half of the asymmetric jaws (Y150 cm, Y2510 cm, X512 cm). One exposure of 50 MU was given from this eld. Then, the superior half of the eld was closed and the inferior half was opened (Y1510 cm, Y250 cm, X520 cm) to give a second exposure using the same MU. Again, measurements were made by varying the position of the Y jaws at the matching area, as in Table 1. Table 1 shows all jaw positions used, designated as settings 14.

Figure 2. Verication lm for one asymmetric jaw to be set exactly at the beam central axis using a collimator rotation of 180 .

Study 2: Assessment of the matching area dose in the three-beam split eld technique using matched collimators
This study was undertaken using 6 MV from the 600C Linac. A lower set of asymmetric collimators was used. Before proceeding with this study, we veried that one of the asymmetric jaws was set exactly at the beam central axis (Figure 2). We adopted the set-up used by Saw et al [10]. An asymmetric eld size of 10 cm6 10 cm was set up with the jaw edge positioned at approximately the beam central axis. A lm placed perpendicular to the beam at a common isocenter and at a depth of the maximum dose was irradiated. For this irradiation, the collimator angle was set at 90 . The collimator was then rotated 180 and the same lm was irradiated
Table 1. Jaw positions (mm) for both the three-beam and two-beam split eld techniques. The superior eld is set with gantry 90 and 270 in the three-beam technique but 0 in the two-beam technique. The inferior eld is set with gantry 0 in both techniques Setting Superior eld Y1 1 2 3 4 0 0 +1 +1 Y2 100 100 100 100 Inferior eld Y1 100 100 100 100 Y2 0 +1 0 +1

again. The lm was developed and dosimetry was examined at the junction. If the jaw edge was set precisely at the beam central axis, the dose prole across the two elds would be uniform. If the dose prole was not uniform, the jaw edge was moved by multiple repetitions of opening and returning to the zero position, and the process was repeated until a homogeneous dose was obtained. After the above initial alignment, the asymmetric jaw at the beam central axis was not moved. Then, the monoisocentric three-beam split elds were set. The superior half was irradiated laterally at gantry angles of 90 and 270 . The collimator was rotated 180 to dene the inferior eld. Then, the lm was irradiated anteriorly at a gantry angle of 0 . Note, in all three beams, the inferior asymmetric jaw was left at the beam central axis and the other jaws were set to dene the elds. To conrm reproducibility, we repeated the measurements for the three- and two-beams with overlapped and matched collimators a total of four times over a 1-year period.

Results
Positional accuracy of asymmetric collimators
The accuracy of each independent jaw as determined by visualizing the light eld at the
The British Journal of Radiology, May 2002

Y1 and Y2, the upper pair of jaws closest to the beam source.

430

Abutment using asymmetrical collimators

zero position was within 1 mm for all Y jaws on both machines. In the case of the 2100C machine, with multiple repetitions of opening and returning the jaw to the zero position, the location of the jaw as judged by the light eld did not vary, nor did the lm dosimetry vary with multiple repetitions of opening and closing of the jaw. An underdose of 25% was always produced at the matching area. This was not the case for the 600C machine. Multiple repetitions of opening and returning the jaw to the zero position always produced different jaw locations (within 1 mm). Again, the lm dosimetry revealed a wide range of dose variations, including a homogeneous dose or large inhomogeneities of up to 35% above or below the prescribed dose.

Study 1: Assessment of the matching area dose in the three-beam split eld technique using overlapped collimators
In the three-beam split eld technique, when setting Y1 and Y2 at the zero position the jaws overlapped, resulting in an underdose at the matching area (Figure 3). Moving the jaw edge by +1 mm either from the superior or the inferior side, a homogeneous dose equal to the prescribed dose was observed (Figures 4 and 5). When the jaw edge was moved +1 mm away from both sides of the abutting elds, an overdose was produced indicating excessive collimator offsetting or eld overlap (Figure 6). Dose prole measurements showed an underdose of about 15%, a homogeneous dose and an overdose of more than 10% of the prescribed dose for settings 1, 2 and 3, and 4, respectively (Figure 7). For the two-beam split eld technique, Figure 8 shows the dose proles across the junction. Although a qualitatively similar dose distribution

Figure 4. Film dosimetry of the three-beam split eld (setting 2) shows a nearly uniform dose at the junction after offsetting the collimator +1 mm from the supraclavicular side.

Figure 5. Film dosimetry of the three-beam split eld (setting 3) shows the same result as in Figure 2 after offsetting the collimator +1 mm from the lateral side.

Figure 3. Film dosimetry of the three-beam split eld (setting 1) shows consistent underdose at the junction owing to consistent collimator overlap at the zero position. The British Journal of Radiology, May 2002

Figure 6. Film dosimetry of the three-beam split eld (setting 4) shows overdose at the junction after offsetting the collimator +1 mm from both sides, indicating too much collimator offsetting. 431

K Abdel-Hakim, T Nishimura, M Takai et al

Figure 7. Proles of the three-beam split eld show approximately 15% underdose (setting 1), nearly uniform dose (settings 2 and 3) and more than 10% overdose (setting 4) for the zero position, +1 mm offsetting from one side and +1 mm offsetting from both sides, respectively.

Figure 9. Film dosimetry of the two-beam split eld with a collimator rotation of 180 , shows a nearly homogeneous dose at the junction.

was observed, from the quantitative point of view, the dosimetric inhomogeneity was more prominent than those obtained from three-beam split elds. With two abutted elds, setting 1 shows a cold spot of about 25% at the matching area, which could be improved only to 8% with a +1 mm collimator offset (settings 2 and 3). An increase of collimator offset to +2 mm (setting 4) produced a dose inhomogeneity of more than 15% at the junction.

junction, abutting three-beam split elds yielded an overdose of about 8% of the prescribed dose (Figure 11). Comparing the dose proles of both techniques (Figures 7 and 8), the same result was obtained using the overlapped collimators. Repeated lm dosimetry exhibited consistent results with maximum variation of less than 1%.

Study 2: Assessment of the matching area dose in the 3-beam split eld technique using matched collimators
The homogeneous matching area dose shown in Figure 9 conrms the position of one of the asymmetric jaws set precisely at the junction. When changing the technique of abutting from twobeam to three-beam split elds, a dose inhomogeneity was produced (Figure 10). Although the two-beam split elds were perfectly abutted with a nearly homogeneous dose prole across the

Figure 10. Film dosimetry of the three-beam split eld with matched collimators, shows an overdose at the junction.

Figure 8. Proles of the two-beam split eld show approximately 25% underdose (setting 1), 8% underdose (settings 2 and 3) and more than 15% overdose (setting 4) for the zero position, 1 mm offsetting from one side and 1 mm offsetting from both sides, respectively. 432

Figure 11. Proles of the three-beam split and twobeam split elds with matched collimators. Note an 8% overdose at the junction owing to variation in the abutting technique. The British Journal of Radiology, May 2002

Abutment using asymmetrical collimators

Discussion
Abutting elds are routinely implemented in head and neck treatments in our department. The availability of asymmetric collimators has facilitated the abutment of split elds. However, because of inherent mechanical and electronic tolerance, it is unavoidable that a junction will not be perfectly abutted. An acceptable dose distribution within +7%/25% across the junction may be obtained if the overdose and underdose cancel each other. This can be achieved in case of an inconsistent collimator misalignment. Practically, digital display tolerance may lead to consistent collimator underlap or overlap although the display indicates perfectly abutted elds. In our clinic, the 2100C linear accelerator with 4 MV energy showed a consistent collimator overlap of approximately 1 mm when abutting elds. It should be noted that 4 MV energy was not deliberately chosen. The focus of this work was to study the consistent collimator overlap, and this linear accelerator incidentally had 4 MV energy. Clinically, no differences in outcome, acute toxicity or late toxicity are discernible in head and neck cancer patients treated with 60Co, 4 MV or 6 MV [14]. In general, collimator alignment inaccuracy is related to independent jaw, gantry and collimator rotation tolerance. Two potential sources of error must be considered in case of consistent collimator overlap. The rst is the weight of the head, which might cause displacement of the beam axis. With the beam pointing down, the weight of the head causes a deection in the gantry. This is seen as displacement in the anterior eld. The second is the display calibration algorithm, which is often adjusted during the linac installation. The adjustment process could lead to consistent collimator overlap or underlap, although the digital display indicates perfectly abutted elds. When using the monoisocentric technique, there is always some doubt about accuracy and uniformity of dose at the junction of the lateral and anterior portals. Consistent collimator overlap, with a resultant underdose of approximately 25%, should make the oncologist concerned about the ability to effectively irradiate all sites of disease. In the International Commission on Radiation Units and Measurements report No 50 [15], the degree of heterogeneity inside the target volume should be kept within +7% and 25% of the prescribed dose. As the matching area is part of the target volume, efforts should be made to smooth out the dose at the junction. Fabrizio et al [11] recommended the use of a penumbra generator to decrease the dosimetric effects of collimator misalignment. Their proposed technique creates an enlarged penumbra in
The British Journal of Radiology, May 2002

both adjacent beam-split-matched elds and then intentionally overlaps them so that the 50% isodose lines line up. However, the use of a penumbra generator has some disadvantages, such as the increase in the volume receiving an inhomogeneity, labour intensive fabrication and mounting, the need for re-entering the treatment room during the treatment, lifting of tray mounted blocks and inaccuracy in tray position. These disadvantages may result in a decrease in treatment efciency. Lees study [12] was intended simply to provide a means of reducing the underdose resulting from a gap between the abutted elds. He suggested using a collimator offset in order to reduce the amount of underdose. However, in his assessment, only two-beam split elds were abutted. In this work, bilateral parallel-opposed elds were orthogonally matched to an anterior eld with a common isocenter typically used in head and neck cancer treatments. We showed that a planning overlap of +1 mm yielded an ideal geometric dose distribution across the matching area with a dose matching the prescription. The technique is easy, reproducible, rapidly set up on a daily basis and free of practical implementation difculties. Using overlapped collimators, the values of three-beam split elds were not consistent with the values obtained when abutting two elds in this study, probably owing to the variation in the abutting technique. In the three-eld technique, two lateral-opposing beams broaden the penumbra in such a way that may or may not resemble the dose gradient of the perpendicular anterior eld depending on the distance between the lateral and anterior elds. Therefore, the effect of the monoisocentric three-beam split eld technique on matching area dose was also investigated using matched collimators. The experiment was performed by initially dening the beam central axis using collimator rotation, which is the standard method of dening the beam central axis. This method of alignment also minimizes the variability of positioning the asymmetric collimator of both elds. Perfectly matched elds were veried by checking the light eld and lm dosimetry. After this preliminary step, the matching area dose for three-beam split elds was assessed using a collimator rotation of 180 . The study revealed an 8% increase in the matching area dose when abutting three-beam split elds after setting the collimator exactly at the beam central axis. Clinically, this nding is important in head and neck cancer treatments. In this situation, the monoisocentric three-beam split eld technique may improve the underdose or worsen the overdose of the matching area, depending on the degree of jaw misalignment. A group of our head and neck
433

K Abdel-Hakim, T Nishimura, M Takai et al

patients is now under treatment using this method and will be subjected to clinical evaluation In summary, current linear accelerator specications for positional accuracy of asymmetric collimators are not rigorous enough to ensure that a clinically acceptable match is produced. Therefore, a collimator offset is a practical solution when an ideal homogeneous dose is desired in head and neck treatments using consistently overlapped collimators. This can be applied to our machine and other similar machines. However, it is necessary to investigate the accuracy and reproducibility of collimator position for any given linear accelerator. The monoisocentric three-beam split elds lead to an increase in the matching area dose by 8% of the prescribed dose in comparison with two-beam split elds. This increase in the matching area dose can be obtained with any linear accelerator and must therefore be taken into consideration in head and neck eld matching.

References
1. Chui C, Mohan R, Fontenia D. Dose computations for asymmetric elds dened by independent jaws. Med Phys 1988;15:925. 2. Khan FM, Gerbi BJ, Deibel FC. Dosimetry of asymmetric X-ray collimators. Med Phys 1986;13: 93641. 3. Losshek DD. Analysis of tissue-maximum ratio/ scatter-maximum ratio model relative to the prediction of tissue maximum ratio in asymmetrically collimated elds. Med Phys 1988;15:67282. 4. Losshek DD, Keller KA. Beam prole generator for asymmetric elds. Med Phys 1988;15:60410. 5. Palta JR, Ayyangar KM, Suntharalingam N. Dosimetry characteristics of a 6 MV photon beam from a linear accelerator with asymmetric collimator jaws. Int J Radiat Oncol Biol Phys 1988;14: 3837.

6. Slessinger ED, Gerber RL, Harm WB, et al. Independent collimator dosimetry for a dual photon energy linear accelerator. Int J Radiat Oncol Biol Phys 1993;27:6817. 7. Stern RL, Rosenthal SA, Doggett EC, et al. Applications of asymmetric collimators of linear accelerators. Med Dosim 1990;20:958. 8. Shon JW, Suh JH, Pohar S. A method for delivering accurate and uniform radiation dosages to the head and neck with asymmetric collimators and a single isocenter. Int J Radiat Oncol Biol Phys 1995;32:80913. 9. Rosenthal DI, McDonough J, Kassaee A. The effect of independent collimator misalignment on the dosimetry of abutted half-beam blocked elds for the treatment of head and neck cancer. Radiother Oncol 1998;49:2738. 10. Saw CB, Krishna KV, Enke CA, et al. Dosimetric evaluation of abutted elds using asymmetric collimators for treatment of head and neck. Int J Radiat Oncol Biol Phys 2000;47:8214. 11. Fabrizio PL, McCullough EC, Foote RL. Decreasing the dosimetric effects of misalignment when using a mono-isocentric technique for irradiation of head and neck cancer. Int J Radiat Oncol Biol Phys 2000;48:162334. 12. Lee PC. Consistent collimator overlaps in eld matching with computer-controlled x-ray collimator. Med Dosim 1997;22:5961. 13. Kutcher GJ, Coia L, Gillin M, et al. Comprehensive QA for radiation oncology: report of AAPM Radiation Therapy Committee Task Group 40. Med Phys 1994;21:581618. 14. Aref A, Berkey BA, Schwade JG, et al. The inuence of beam energy on the outcome of postoperative radiotherapy in head and neck cancer patients: secondary analysis of RTOG 85-03. Int J Radiat Oncol Biol Phys 2000;47:38994. 15. International Commission on Radiation Units and Measurements. Prescribing, recording, and reporting photon beam therepy. Report 50. Washington DC: ICRU, 1993.

434

The British Journal of Radiology, May 2002