0360-3016190 $3.00 t .30 Copyright lb 1990 Pergamon Pres plc
Ini J. Radiatmn Oncology Bid. Phys.. Vol. 18. PP. 1223-1232 Printed in the U.S.A. All rights reserved.
0 Technical Innovations and Notes DESIGN OF METALLIC ELECTRON BEAM CONES FOR AN INTRAOPERATIVE THERAPY LINEAR ACCELERATOR K. R. HOGSTROM, PH.D.,’ A. L. BOYER, PH.D.,’ A. S. SHIU, PH.D.,* T. G. OCHRAN, S. M. KIRSNER, M.S.,2 F. KRISPEL, PH.D.~ AND TYVIN RICH, M.D.’
M.S.,’
‘Departments of Radiation Physics and Clinical Radiotherapy, The University
of Texas M. D. Anderson Cancer Center, Boulevard, Houston, TX 77030; ‘Abington Hospital, Radiation Therapy Department, 1200 Old York Road. Abington, PA 1900 1; and ‘Siemens Medical Laboratories, Inc., 4040 Nelson Ave., Concord, CA 94520
1515 Holcombe
A set of circular collimators and treatment cones from 5 to 12 cm diameter has been designed for an intraoperative accelerator (6-18 MeV) that has an optica1 doeking system. Electron beam scattering theory has been used to minimize their weight while minimizing Ieakage radiation. Both acrylic and brass were evaluated as possible materials; however, because of substantial electron leakage through the lateral tone wal1 for acrylic, we have concluded that 2 mm thick brass walls are more desirahle than acrylic walls. At 18 MeV, isodose measurements beneath the cones showed hot spots as great as 120% for both materials. The placement and dimension of an internal trimmer ring inside the brass tone was studied as a method for reducing the hot spots, and it was found this could only be accomplished at the expense of decreasing coverage of the 90% isodose surface. The effects of 1’ tone misalignment on the dose distribution has been studied and found to generate changes of less than 5% in the dose and 3 mm in position of the 90% isodose surface. In a study of the contribution of the tone and its matching collimator assembly to x-ray room Ieakage, it was noted that although the treatment tone had a negligible contribution, the upper annuli of the upper collimator assembly contributed as much as 80% of the leakage at 16 MeV for the 5cm tone. Electron cones, Electron dosimetry, Linear accelerators, Intraoperative radiotherapy.
INTRODUCTION
designed to deliver electron beams from 6 MeV to 18 MeV and to produce flat beams for field sizes as large as 15 cm in diameter. A number of clinical contraints were imposed upon the design. First, clinical experience suggested a tone length of 25-30 cm is needed to reach tumors deep within the abdomen. Second, each tone should be as light as possible. Third, the collimation system and tone should be easily sterilizable. Fourth, the collimation system should allow only minima1 leakage to the patient. Fifth, misalignment due to the laser docking system should minimally affect the dose distribution and radiation leakage. Sixth, the resulting dose distribution within the patient should be as uniform as possible inside the 90% isodose surface, and the geometrie coverage of the 90% isodose surface should be as large as possible. Seventh, the effect of the collimation system on x-ray leakage to the adjacent rooms should be minimal. This latter constraint is important in a treatment facility where the accelerator is being installed in an existing operating room, as there
This paper describes an approach and concepts in the design of a tone system for collimation in intraoperative electron therapy. A tone, which is a tube with a circular, rectangular, or other regularly shaped cross section through which the electrons pass in route to irradiating the target volume in the patient, is practica1 for this application. The wal1 of the tone shields the patient anatomy outside the tone from primary radiation. The homogeneity of dose to the patient’s target volume and the leakage dose outside the tone depend on the tone design and how it is interfaced to the therapy machine. A number of investigators ( 1,2, 3,9, 11) have designed or modified tone systems that attach directly to the head of the linear accelerator. In the present work, results are reported for a tone system designed for a dedicated intraoperative electron accelerator.* This system is unique in that it uses a laser docking system in which the patient tone is not in direct contact with the treatment head. The machine was
Reprint
Accepted for publication 9 August 1989. * Mevatron-ME, Siemens Medical Systems, CA.
requests
to: Kenneth R. Hogstrom, Ph.D. investigation was supported in part by a research contract with Siemens Medical Laboratories, Inc., and by a grant from the National Cancer Institute, CA-06294.
Acknowledgements-This
1223
Inc.. Concord.
1224
1. J. Radiation Oncology 0 Biology 0 Physics
may be structural limits to the amount of shielding that can be added. We discuss a number of design techniques, including determination of the thickness of circular trimmers (annuli) and the wal1 thickness of cones; determination of the geometry of annuli; and evaluation of internal blocking ring geometry. Two sets of circular cones with intemal diameters of 5 cm to 12 cm in 1-cm steps were specified, one with the lip of the tone perpendicular to the beam, and one with a 22.5” angle relative to the perpendicular plane. Figure 1 illustrates the location and geometry of the collimators. The upper collimator assembly consists of two annuli (upper and lower) fixed to the head of the machine. The clinical beam is defined by a third annulus (tone annulus) attached to the treatment tone and detached from the machine. lt is aligned with the treatment head via a laser docking system (Fig. 2), which employs four pairs of laser beams. When each pair (a line and a point) intersects to form four points every 90” along a 15.2 cm diameter circle, the treatment cone’s lateral position, distance from the treatment head, and angle are correct. Each laser spot is approximately 1 mm in diameter, and experience has shown agreement in alignment of the spots as smal1 as 1 mm are easily achievable. Therefore, alignment should be within 1 mm in the lateral dimension, 0.7 mm along centra1 axis, and 0.5” in angulation.
May 1990, Volume 18. Number 5 Laser Beams (8)
Fig. 2. Laser alignment system. Eight laser beams, four lines and four points, intersect on four points that lie on the circumference of a circle scribed on the alignment plate when the treatment tone is in proper alignment.
METHODS
AND
MATERIALS
Design theqv
Treatment
Head
UpperAnnulus
UPPer Collimating
Lower
1
Cone
Annulus
Assembly
Annulus
Fig. 1. Schematic of collimating system of the electron beam. The upper two annuli, forming the upper collimator assembly, are rigidly attached to the treatment head. There is a 15-cm air gap between it and the treatment tone.
In determining the size and shape of the electron collimator assembly and treatment tone, two types of electron leakage were considered: the transmission of electrons through the collimator material and the scattering of electrons outside the outer edge of the annuli. Collimutor material and thickness. The upper collimator annuli and the annulus on top of the treatment tone are brass. Brass was selected because of its high density, durability, machinability, and medium atomic number. It was felt that excessive bremsstrahlung production in a higher Z material would exacerbate room-to-room x-ray leakage. On the other hand, the low density of lower Z materials would require substantially thicker collimating annuli, perhaps leading to needless electron leakage from scatter off the annuli walls in the upper collimator assembly. Transmission measurements in brass (5) have shown that 12.7 mm is sufficiently thick to stop I8-MeV electrons. Both brass and acrylic (polymethylmethacrylate) were evaluated for the treatment tone. Acrylic offers the advantage of being transparent and has been frequently used by others for intraoperative electron therapy. Brass is durable, can be chrome plated, and flash sterilized. Its walls can be made thinner than acrylic, permitting better tumor coverage in tight-fitting anatomical situations. As a minimum, the tone wal1 thickness should be such that the thickness along a diagonal ray from the source is at least
Metallic cones for intraoperative electron therapy 0 K.
Electron Beam
the maximum range (R,,,) of a broad electron beam (in brass approximately 12.7 mm at IS MeV). The maximum diagonal thickness occurs for the largest field diameter and that portion of the tone closest to the source. When the source-to-surface distance is 100 cm, a 15-cm diameter, 30-cm long tone has a maximum diagonal wal1 thickness given by t,_,, = R,,,
* sin 8,,,
= 0.1 - R,,,.
2= fJx
$ (
‘T”.(Zi - Zi_1)2*Zi/6, i 81r
Air
(1)
Hence, a wal1 thickness of approximately 1.3 mm for brass and 8.8 mm for acrylic is required. The prototype treatment cones were only 25-cm long, and 1-mm brass wal1 thickness was selected. On the other hand, 8.8 mm acrylic is quite thick; therefore, 5 mm was selected for study based on the recommendations of McCullough and Biggs (10) of 3-5 mm. As noted in the results, greater wal1 thickness is required to control leakage through the side of the tone. Geometr)! o/‘collimating annuli. The size and location of the collimating annuli were determined using the clinical constraints described earlier and the theory of electron transport through air. As electrons pass through air they undergo multiple Coulomb scattering, which results in a lateral spreading of the electron beam. Lateral electron distributions in air can be calculated using Fermi-Eyges multiple scattering theory (7). In Figure 1, the first two annuli, the upper annulus and the lower annulus. form an upper collimator assembly that is attached to the treatment head. The third annulus is an integral part of the treatment tone itself. The lateral dimensions of the collimator annuli were selected to minimize weight and leakage. Figure 3 illustrates the principle of the collimator design (5). The penumbra generated by an upstream collimator must be shielded by a downstream collimator. For a collimator edge perpendicular to the centra1 axis, the outer edge of the downstream collimator should extend to at least 3.1 times cX outside the projected edge of the upstream collimator for the leakage due to electron scatter to be less than 2%; the inner edge should lie 1.3-1.6 times u, inside the projected edge to ensure uniformity of electron fluence within the aperture. cX is the root mean square value of the lateral spatial distribution of an electron pencil beam originating at the upstream collimator and arriving at the downstream collimator. It is derived from fundamental pencil beam equations (6. 7). resulting in
(2)
where Zi is the distance from the electron source to the top of the downstream collimator and Zi-1 is the distance from the source to the bottom of the upstream collimator. At sea level, the linear angular scattering power in air for the most probable incident energy, E,.o, equaling 6 MeV is approximated by (4)
1225
R. HOCSTROM et UI
I/
Fig. 3. Electron collimator design concept. The downstream annulus is designed to intercept the penumbra due to the upstream collimator (from the 2% profile intensity to the 95%’profile intensity).
= 2.3 - lom4 radian’/cm.
(3)
For calculation purposes. 6 MeV was selected because electrons scatter the most at the lowest energy. Our design theory strictly applies to rectangular fields. Rut when it is applied to circular fields, it results in an upper limit for leakage. Also for treatment facilities wel1 above sea level, where air scattering is less, this should again yield an upper limit. The upper and lower annuli diagrammed in Figure 1 are rigidly attached to the treatment head. The upper annulus is just below the collimation in the treatment head, and its outer edge extends outside the aperture in the treatment head so that no electrons can escape. The penumbra tast by the inner edge of the upper annulus is intercepted by the lower annulus. Similarly, the tone annulus is designed to intercept the penumbra tast by the lower annulus of the upper collimator assembly.
Misalignment
in plane perpendicular to centra1 axis.
Because the collimators were designed to be used in a “soft” docking system, translation errors in the alignment procedure must be allowed. It was felt that a 1-mm translation accuracy should be easily achievable; nonetheless, a conservative *5 mm misalignment criteria was incorporated into the final design. This means that the electron beam fluence, incident on the tone annulus should be uniform for an additiona10.5 cm in radius outside a given annulus aperture. For this reason the outer radius of the tone annulus was increased by 1 cm in the final design (not in the prototype) to prevent leakage, 0.5 cm to account for the broader diameter of the incident fluence,
1226
1. J. Radiation Oncology 0 Biology 0 Physics
and 0.5 cm to account for misalignment. This design feature adds additional weight to the tone, for example, 0.4 kg for the 9-cm diameter tone. Evaluation criteria. Al1 standard dosimetry quantities for electron beams are of interest: (a) surface dose, therapeutic depth, dose gradient, and bremsstrahlung for the depth-dose curve, (b) beam flatness, (c) the therapeutic isodose (90%) coverage; and (d) collimator leakage. For intraoperative therapy, not only is leakage measured in a phantom at the end of the tone important but so is leakage through the side of the tone because of healthy tissues that might be adjacent to the tone wall. Methods of measurement A variety of measurement conditions were used on this project because of the ongoing development of both the tone system and the accelerator. Data were taken at both Siemens Medical Laboratories, Inc. (SML) and M. D. Anderson Cancer Center (MDACC). Initial measurements were made at the SML factory on the intraoperative electron accelerator.* Five electron energies, 6, 9, 12, 15, and 18 MeV, were available. For the 18-MeV beam, a most probable energy of 17.4 MeV was the maximum achievable from the magnetrondriven accelerator. The prototype collimation system used 25 cm long brass and acrylic cones. Ionization measurements were made using a 0.14 cm3 cylindrical air ionization chamber (0.4 cm intemal diameter) and electrometer,+ which was scanned by an analog linear scanner. Off-axis stans were made both in air and in a water phantom. Ionization readings were normalized to 100% on centra1 axis. Film* measurements were made in a “solid-water” phantom. The film phantom had a cassette that accepted bare film and ensured that the film’s edge was precisely aligned (kO.1 mm) with the phantom surface. Film was exposed edge-on both beneath the tone in a plane containing centra1 axis and in a plane perpendicular to the tone wal1 and parallel to centra1 axis. Exposures were controlled so that the film optica1 density (O.D.) was within the linear portion of the dose-response curve (O.D. < 2). Film was developed by an automatie processor, and a nonirradiated film was used for determining background. The films were scanned at M. D. Anderson using a homemade scanning densitometer. Subsequent measurements were made using the 20 MeV beam and 7 cm tone assembly on a conventional radiotherapy accelerator* at M. D. Anderson to study the effect of plating a ring inside the tone. These data were taken with the M. D. Anderson data acquisition system,
* Mevatron-ME, Siemens Medical Systems, Creek, CA. + Capintec, Ramsey, NJ. * XTL-2 film, Kodak, Rochester, NY. p Mevatron KD, Siemens Medical Systems, Creek, CA.
Inc.,
Walnut
Inc.,
Walnut
May 1990, Volume 18, Number 5
which uses a 0.1 cm3 PTW air-ionization chamber and a PTW water phantom and scanner.** The charge was integrated by a computer-controlled electrometer.tt The entire system was controlled by a microprocessor*+ with software developed in house. The raw data was transferred into dose using a (3/4) radius displacement correction, a perturbation factor for the cylindrical chamber, and CE to convert the exposure in air to the dose in water. Following preliminary data on design of the internal ring, measurements were made at SML for the 7, 9, and 12-cm cones at energies of 6, 12, and 18 MeV, varying the position of the ring at 5, 10, and 15 cm above the bottom lip of the tone. The optimal position was determined to be 10 cm, and only samples of that data are reported here. Isoionization measurements were made using a 0.14 cm3 ionization chamber +in a Wellhofer water phantom and beam scanning system.@ The final measurements of tone leakage and misalignment studies were made at MDACC after installation of the accelerator. During installation it was decided to replace the 18 MeV beam with a 16 MeV beam so as to prolong the life and dependability of the magnetron. Isodose measurements were made by combining centra1 axis depthdose data, measured using the M. D. Anderson data acquisition system with off-axis profiles measured using film. RESULTS Collimator leakage In preliminary studies, collimator leakage was evaluated both (a) downstream and outside the edges of the annuli (assumed due to scatter and most significant at the lowest energy, 6 MeV) and (b) through the annuli and tone (assumed due to transmission and most significant at the highest energy, 18 MeV). Figure 4 is an in-air radial, relative ionization profile in a plane 1 cm below the lower annulus of the upper collimator assembly used with the 7-cm diameter tone. The low relative ionization under the lower annuli verifies that al1 primary electrons are stopped in the brass. Outside the outer edge of the lower annulus, the leakage radiation is less than 2%, and the beam intensity within the aperture is sufficiently uniform, verifying the proper dimensions of the upper collimator assembly as designed from scatter theory. In Figure 5, the in-air, radial, relative ionization profiles downstream of the upper collimator assembly at the position of the top of the treatment tone are plotted for the upper collimator assembly used with the 7-cm diameter tone. At both 6 MeV and 18 MeV, the annulus (prototype)
** Nuclear Associates, Carle Place, NY. ++Model 6 17, Keithley, Cleveland, OH. #* DEC LSI- 11/02. @ Frank Barker Associates, Pequannock,
NJ.
Metallic cones for intraoperative electron therapy 0 K. R.
-
18 MeV
---
8 MeV
HOGSTROM ef al.
1227
is removed in the final design as the aperture of the upper collimator assembly is increased to move the penumbra an additional 0.5 cm away from centra1 axis. The radial width of the tone annulus was increased 1 cm, 0.5 cm to account for the increase in the diameter of the upper aperture and 0.5 cm to allow for misalignment of the tone laterally. Cone leakage Previous tone designs have primarily used acrylic with walls as thin as 3 mm for 18 MeV electron beams (9). In
Lower Annulus of Upper Collimator Assembly
____-* 0 Radial Distance (cm)
-5
Fig. 4. Comparison of in-air radial ionization profiles for 6-MeV and 18-MeV beams measured 1 cm below the upper annulus. Hashed rectangles indicate the projection of the edges of the
lower annulus. on top of the 7-cm diameter
tone has been positioned to properly intercept the penumbra. Leakage outside the annulus is smal1 and believed mostly to be X rays. Inside the annulus, the beam has good uniformity at 18 MeV; however, at 6 MeV, the relative ionization is below 90% in a region extending from the edge to 0.5 cm inside the edge of the tone annulus. This region of nonuniformity 100
-Y \\
/--,’
: :
:
I I
z s f .s 9 t 3
-
18 MeV
---
8 MeV
Bc
this case, McCullough et al.‘s radial profile data in a water phantom below the tone show leakage in excess of 10% outside the penumbra at 18 MeV for the larger field sizes. We believe this leakage was probably due to transmission leakage through the side of the cones, which could result in substantial leakage at the outer surface of the cones. Fraass et al. (3) have shown that stainless steel wrapped around cones with 6-mm thick acrylic walls is necessary to eliminate leakage for a 20-MeV beam. However, neither of these studies measured leakage dose to tissue that may be adjacent to the outside walls of the collimator. To measure this effect, we exposed films in a solid-water phantom placed perpendicular to the outer surface of the prototype cones and parallel to the centra1 axis. Results of these measurements are shown for a 1-mm thick brass tone and a 5-mm thick acrylic tone in Figures 6 and 7, respectively. Both figures show electron leakage through the walls, which is greatest at the upper portion of the tone, where the electrons have the greatest angle of incidence. Note that the film lateral to the cones was exposed by delivering 10 times the number of monitor units used to expose the film in the treatment field. For the brass tone, the leakage is as great as 8% of the maximum dose on centra1 axis, and for the acrylic tone as large as 16%. The dose fall-off laterally was measured to be 50% in approximately 1.0 gm/cm2 of plastic phantom material. Based on these data, it was decided to increase the thickness of the walls by that equivalent amount. This would make the acrylic cones prohibitively thick (1 cm), but would only increase the brass cones by about another 1 mm.
50
I I I I
i,I
I\
:
\ Annulus on Top of : Treatment Cone !
-4& \ ;;
-5
0 Radial Distance km)
Fig. 5. Comparison of in-air radial ionization profiles for 6-MeV and 18-MeV beams measured at the loeation of the tone annulus. Hashed reetangles indicate the position of the edges of the tone annulus.
Dose distributions The two main features evaluated from the planar dose distribution were coverage of the 90% isodose contour (i.e., the treatment volume) and beam uniformity within that contour. Initial measurements clearly showed that both acrylic and brass cones had scatter off the tone walls resulting in rims of increased dose just inside the field edges at superhcial depths. This is illustrated in the radial, relative dose plot at a depth in water of 0.9 cm in Figure 8 for a 7-cm diameter tone at 18 MeV. Note the lower leakage outside the field of the brass tone (consistent with data of Figs. 6, 7) and that both cones exhibited a 20% hot spot along the rim. At 6 MeV such a comparison again yielded almost identical curves for acrylic and brass,
1228
1. J. Radiation Oncology ??Biology 0 Physics
May 1990, Volume 18, Numher 5
errors might cause significant changes in the dose distributions. For determination of ring geometry, a 20-MeV electron beam+ was used (an I8-MeV beam was not available at the time of testing) because the higher energy exhibited the greatest effect of tone scattering. An intermediate ring position, 10 cm above the bottom of the tone, was selected for the purpose of removing electrons streaming from the upper portion of the tone and decreasing the number of electrons striking the lower portion of the tone. The brass ring was 1.27 cm thick in the direction of the beam; its radial thickness was varied. Figure 9 illustrates isodose plots for use of no ring and rings with radial thicknesses of 0.25 cm and 0.125 cm. The wider ring reduced the hot spot of the rim, but sacrificed coverage of the 90% isodose line. It was originally decided by the radiotherapists that a 10% hot spot ( 110% isodose) would be clinically acceptable, so as not to lose coverage of the 90% isodose. A revised set of cones was then designed and constructed; these were 30 cm long and had walls 1.8-mm
-
7cm
I
-3/16*‘(6mm) Fig. 6. Demonstration of electron leakage through the wal1 of the brass tone for an 18-MeV beam. Note that the leakage radiation is greatest near the top of the tone.
but with a hot spot of only 5% and leakage of less than 1%. As the dose uniformity inside the tone was the same for the two types of material and the brass tone had significantly less leakage than the acrylic tone, we decided to use brass cones in the final design. Two methods of achieving beam uniformity have been previously studied: (a) adjustment of x-ray jaw width ( 1, 3, 11) to adjust the number of electrons striking the wal1 of tone, and (b) insertion of a smal1 collimation ring inside the tone (8). One problem with the first method is that the x-ray jaws form a square field for the circular cones so that beam uniformity on the major axis does not necessarily result in beam uniformity along the diagonals. For the machine we used, this is not a problem, as the circular annuli of the upper collimator assembly define the electron beam incident on the tone and can be adjusted accordingly. However, this method was abandoned and the second method employed because the tone is not rigidly attached to the treatment head and smal1 alignment
rm (5 YU En -2 Film
Fig. 7. Demonstration of electron leakage through the wal1 of an acrylic tone for an 18-MeV beam. Note that the leakage is greater than that of the brass tone in Figure 6.
Metallic cones
for
intraoperative
-
Bram
---
Acrylic
electron therapy ??K.
R. HOGSTROM efal.
1229
CR
I ___---
J I
1
-5
I 0
\ I
Na Ring
5
Radial Distance (cm)
Fig. 8. Comparison of radial-dose profiles in a water phantom at 0.9 cm depth for the acrylic and the brass cones with the 18MeV beam and 7-cm diameter tone. Note the 20% hot spot just inside the penumbra and the leakage just outside the penumbra.
thick for the top 20-cm length and 1.0 mm for the final 10 cm. The internal ring was placed 10 cm from the exit end of the tone. In light of the results shown in Figure 9, the ring radial thickness for each tone was adjusted so that a ray from the source grazing the inner edge of the lower surface of the ring would strike the end of the tone 6.7 cm above the bottom (e.g., 0.125-cm ring for a 7-cm tone). Isoionization curves were subsequently measured at 6, 12, and 18 MeV for the 5-, 9-, and 12-cm diameter cones at the SML factory. The resulting isoionization curve for the 12-MeV beam produced by the 9-cm tone is shown with the ring 10 cm above the bottom lip in Figure 10. Only isoionization contours less than 90% are shown when no ring was used. In summary, the ring has the effect of(a) reducing the hot spot, (b) decreasing slightly the coverage of the 90% isodose, and (c) broadening the lO%-30% isodose lines at depths less than d,,, . The reason for the latter is not fully understood.
Fig. 9. Comparison of isodose curves for varying ring radial thickness for a 20-MeV beam and 7-cm brass tone. The ring is located 10 cm above the bottom lip of the tone. The dashed curve indicates the position of the 90%’isodose in the absente of the ring.
Measurements were made at 6 MeV and the 7-cm tone misaligned by 1 mm, 2 mm, study errors in lateral translation. Based on surements in a plane containing centra1 axis, -
Rins ---NO
18 MeV with and 5 mm to isodose meano differente
Rins
Cone misalignment Because of the potential for tone misalignment when using a laser docking system, the dose distributions were measured with the cones intentionally misaligned. As stated earlier, we believe it is possible and practica1 to achieve alignment of the treatment tone with position errors less than 1 mm in lateral offsets and angles less than 0.5”.
10
I
I
I
-5
0
5
Radial Distance km)
Fig. 10. Isoionization curves for a 9-cm tone with a ring (solid lines) and without a ring (dashed lines) for 12 MeV electrons. Isodose curves above 90% are not indicated for the latter.
1230
1. J. Radiation Oncology 0 Biology 0 Physics
was detected within the accuracy of measurement (& l%, f 1 mm). This was not surprising because the cones were designed to permit a 5-mm error in lateral alignment. On the other hand, we were concerned about angulation errors because of the obvious sensitivity of the dose distributions to scattering off the inside walls of the collimator. Measurements were made at 6 MeV and 15 MeV with the 5-cm and 12-cm cones misaligned by O”, 0.5”, 1.O”, and 1.5”. Isodose measurements in a plane containing centra1 axis for the 5-cm tone using the 16 MeV beam are compared for the 0” and 1.O”misalignment in Figure ll. On the side farthest from the source (-), there is no significant change in the 90% isodose, and within the treatment volume (90% isodose for O”), an increase of O5% in dose is observed. On the side closest to the source, the 90% isodose line lies 0-3 mm inside the treatment volume, and within the treatment volume, a decrease of 0-5% in dose was observed. This type of effect was observed for the 12-cm tone and at 6 MeV. Studies at 0.5” showed the differences to be approximately 50% less and at 1.5” approximately 50% more. X-ray production by cones
The goal of the overall project was to develop an intraoperative radiotherapy suite within a conventional surgical suite using a machine that had only electron beam capability. Radiation shielding for the machine was necessary only for secondary x-ray production in the treat-
Radiel
Distance
(cm)
Aligned (Oo tilt)
Misaligned (le
May 1990, Volume 18, Number 5
ment head, the patient, and the collimating system. A set of measurements was made to determine the relative contribution of the collimating system to x-ray leakage. Measurements of x-ray leakage in the room directly under the machine were made for each of five energies for the electron beam directed downward (0”). These measurements were made at point A, located on centra1 axis 3.3 m below isocenter, and at point B, located at the same centra1 axis distance, but 3 m lateral to centra1 axis in the gun-target plane. Point A was selected as being directly under 20 cm of lead shielding in the floor, which shields the primary beam. Point B is selected to lie wel1 outside the primary beam and hence outside the lead shielding of the floor. The measurements were made for the 5-cm and 12-cm diameter cones at the highest energy achievable in the final installation, 16 MeV. Measurements were made (a) with the full collimation system in place, (b) with the tone removed, (c) with the tone and the lower annuli of the upper collimator assembly removed, and (d) with both the tone and the upper collimator assembly removed. In al1 cases, the electron beam was absorbed by a 25 cm deep water phantom at the 100 cm SSD. The results of these measurements are listed in Table 1. Removal of the tone resulted in only a modest decrease in the leakage radiation, being more significant for the higher energy, larger tone, and point B outside the beam. Qualitatively, this was expected, as x-ray production increases with energy, the larger tone has more surface area, and there is little attenuation of radiation produced in the tone by the floor at point B. The magnitude of leakage produced in the tone is insignificant from a safety perspective. Removal of the lower annulus of the upper collimator assembly has an even less significant effect. On the other hand, a significant reduction in leakage was observed with the removal of the upper annulus, indicating it was a major source of leakage. As expected qualitatively, the reduction in leakage was most significant for the higher energy, smaller tone, and point B outside the beam. In this case, the smaller tone has the largest upper annulus in the upper collimator assembly, hence producing the most X rays. At 16 MeV for the 5-cm tone and at point B, a decrease of 83% is observed, which is significant from a safety perspective. The upper annulus appears to be the major contributor to leakage radiation; therefore, efforts should be directed toward replacing al1or part of the upper aperture with a lower Z material such as aluminum (p = 2.7), beryllium (p = 1.8), or encased graphite (p = 2.3).
tilt)
SUMMARY
Fig. ll. Comparison of isodose curves for the 5-cm diameter tone at 16 MeV, for the tone properly aligned with those for the tone angulated 1’. The dashed curve indicates the position of the 90% isodose when the tone is aligned.
It has been demonstrated that the lateral dimensions of electron collimators (annuli) can be determined using multiple scattering theory at the lowest electron energy and that the thickness can be determined from the maximum range, R,,, measured for electrons in the collimator materials at the highest energy. On this basis, the
1231
Metallic cones for intraoperative electron therapy 0 K. R. HOGSTROM etal.
Table 1. Influence of collimation Collimation present*
Location 12-cm tone Point A
Point B
5-cm tone Point A
Point B
6 MeV
on x-ray leakage (al1 data in mR - h-‘) 9 MeV
12 MeV
15 MeV
16 MeV
1 2 3 4
1.9 1.9 1.9 1.8
5.5 5.2 5.2 4.8
11.0 9.4 9.6 8.4
18.0 17.5 17.0 14.5
21.0 18.0 18.0 15.0
1 2 3 4
4.0 3.8 3.6 2.0
11.0
10.5 10.0 4.2
21.0 20.0 19.0 6.0
42.0 34.0 33.0 9.4
42.0 37.0 35.0 9.4
1 2 3 4
2.5 2.4 2.3 1.8
6.4 6.2 6.2 4.6
12.0 12.0 12.0 8.2
23.0 21.0 21.0 15.0
24.0 23.0 23.0 15.0
1 2 3 4
5.4 5.4 5.2 2.0
15.0 15.0 15.0 4.4
30.0 28.0 28.0 6.0
50.0 49.0 48.0 9.4
56.0 54.0 54.0 9.4
* 1 = upper aperture and lower aperture of upper collimator assembly and tone; 2 = upper aperture and lower aperture of upper collimator assembly; 3 = upper aperture of upper collimator assembly; 4 = upper collimator assembly and tone removed.
collimation
system
for a set of circular
brass cones has
been designed for a radiotherapy machine* specially designed for intraoperative electron therapy. Note, that careful attention must be paid to the leakage through the side of a treatment tone, primarily due to electron scatter. At 18 MeV, leakage through 5-mm acrylic cones was as much as 16% and through brass cones with 1-mm radial thickness as much as 8%. In both cases, the radial thickness must be approximately doubled to reduce the electron leakage by half. Hot spots in the isodose distributions for the radial planes, as great as 120%, were observed due to scatter off the inside wal1 of the tone. In an effort to remove this rim of increased dose, an internal ring was placed inside the tone 10 cm upstream of the bottom lip of the tone. Results showed that a radial width, whose inside edge projects 6.7 cm from the bottom lip of the tone, reduces the hot spot to approximately 110%. By increasing the radial thickness of the ring, the hot spot can be further reduced; however, each reduction in the hot spot is accompanied by a corresponding decrease in the coverage of the 90% isodose surface. Therefore, any evaluation of clinical advantages of this ring must weigh more uniform dose against poorer coverage of the treatment volume at depth. Because the treatment tone is docked using a laser alignment system, the tone may be mispositioned. A study of these errors showed that a 5-mm translation error had no significant effect on the relative dose distribution. Translation errors greater than 2 mm are not expected. Data show that a 1o alignment error increases the dose by as much as 5% on the side of the tone farthest from
the source and decreases the dose by as much as 5% on the opposite side. Rotation errors greater than 1’ are not expected, so this is not expected to be a clinical problem. A study of radiation leakage to the room below revealed that the electron collimation system was a major source of x-ray leakage. The electron tone assembly and the lower annuli of the upper collimator assembly contributed negligibly to the leakage. On the other hand, the upper annulus of the upper collimator assembly contributed as much as 80% of the leakage for the smal1 field sizes at the highest energy, 16 MeV. In other words, that annulus is intercepting a significant portion of the beam and is serving as a secondary x-ray source. It should be possible to reduce this leakage by replacing al1 or part of the upper annuli, made of brass, with one of a lower Z material. During the course of these studies, a computer program that designs electron treatment cones was developed. This software was used to design a circular set of cones with diameters from 5 to 12 cm in 1-cm steps and the following features: 30-cm length, brass composition, 1.78 mm wal1 thickness the upper 20 cm, 1 mm the lower 10 cm, allowance for a 0.5-cm misalignment, and no internal ring. The radiotherapist’s decision to remove the ring was based on the philosophy that the benefit of the increased coverage of the 90% isodose line was worth the smal1 volume of increased dose. The resulting cones weigh 1.9 kg for the 7 cm diameter, 2.4 kg for a 9 cm diameter, and 3.0 kg for the 12-cm diameter tone. The collimation system can produce good electron dose distributions reliably and safely during intraoperative radiotherapy and appears suitable for clinical use.
232
1. J. Radiation Oncology 0 Biology 0 Physics
May 1990. Volume 18, Number 5
REFERENCES 1. Bagne, F. R.; Samsami, N.; Dobelbower, R. R. Radiation contamination and leakage assessment of intraoperative electron applicators. Med. Phys. 15530-537: 1988. 2. Biggs, P. J.; Epp, E. R.; Ling. C. C.: Novack, D. H.; Michaels. H. B. Dosimetry, field shaping and other considerations for intra-operative electron therapy. Int. J. Radiat. Oncol. Biol. Phys. 7:875-884; 198 1. 3. Fraass, B. A.; Miller, R. W.; Kinsella, T. J.; Sindelar, W. F.; Harrington, F. S.; Yeakel, K.: Van De Geihn, J.; Glatstein, E. Intraoperative radiation therapy at the National Cancer Institute: Technical innovations and dosimetry. Int. J. Radiat. Oncol. Biol. Phys. 1 1: 1299- 13 11; 1985. 4. Hogstrom, K. R.; Kurup, R. J.; Shiu, A. S.; Starkschall, G. A two-dimensional pencil-beam algorithm for calculation of arc electron dose distributions. Phys. Med. Biol. 34:3 I5341; 1989. 5. Hogstrom, K. R.; Meyer, J. A.; Melson, R. Variable electron collimator for the Mevatron 77: design and dosimetry. In: Proceedings of the 1985 Mevatron Users Conference. Iselin, N.J.: Siemens Medical Systems, Inc.; 1986:25 1-276.
6. Hogstrom, K. R.; Mills, M. D.; Almond. P. R. Electron beam dose calculations. Phys. Med. Biel. 26:445-459; 198 1. 7. Huizenga, H.: Storchi, P. R. M. The in-air scattering of clinical electron beams as produced by accelerators with scanning beams and diaphragm collimators. Phys. Med. Biol. 32:101 1-1029; 1987. 8. Kao, M.; Lanzl, L.: Rozenfeld, M.: Pagnamenta. A. Dose uniformity of the intraoperative radiation therapy tone (Abstract). Med. Phys. 13:605: 1986. 9. McCullough, E. C.; Anderson, J. A. The dosimetric properties of an applicator system for intraoperative electronbeam therapy utilizing a Clinac- 18 accelerator. Med. Phys. 9:261-268; 1982. 10. McCullough, E. C.; Biggs, P. Physical aspects of intraoperative electron beam radiation. Curr. Probl. Cancer 7:2430: 1983. 11 Ochran, T. G.; Schabinger, P. R. Clinical application and dosimetric evaluation of a state-of-the-art intraoperative tone system. In: Proceedings of the 1987 Therapy Users Conference. Iselin, NJ: Siemens Medical Systems, Inc.: 1987: 2 15-240.
Ini J. Radiatmn Oncology Bid. Phys.. Vol. 18. PP. 1223-1232 Printed in the U.S.A. All rights reserved.
0 Technical Innovations and Notes DESIGN OF METALLIC ELECTRON BEAM CONES FOR AN INTRAOPERATIVE THERAPY LINEAR ACCELERATOR K. R. HOGSTROM, PH.D.,’ A. L. BOYER, PH.D.,’ A. S. SHIU, PH.D.,* T. G. OCHRAN, S. M. KIRSNER, M.S.,2 F. KRISPEL, PH.D.~ AND TYVIN RICH, M.D.’
M.S.,’
‘Departments of Radiation Physics and Clinical Radiotherapy, The University
of Texas M. D. Anderson Cancer Center, Boulevard, Houston, TX 77030; ‘Abington Hospital, Radiation Therapy Department, 1200 Old York Road. Abington, PA 1900 1; and ‘Siemens Medical Laboratories, Inc., 4040 Nelson Ave., Concord, CA 94520
1515 Holcombe
A set of circular collimators and treatment cones from 5 to 12 cm diameter has been designed for an intraoperative accelerator (6-18 MeV) that has an optica1 doeking system. Electron beam scattering theory has been used to minimize their weight while minimizing Ieakage radiation. Both acrylic and brass were evaluated as possible materials; however, because of substantial electron leakage through the lateral tone wal1 for acrylic, we have concluded that 2 mm thick brass walls are more desirahle than acrylic walls. At 18 MeV, isodose measurements beneath the cones showed hot spots as great as 120% for both materials. The placement and dimension of an internal trimmer ring inside the brass tone was studied as a method for reducing the hot spots, and it was found this could only be accomplished at the expense of decreasing coverage of the 90% isodose surface. The effects of 1’ tone misalignment on the dose distribution has been studied and found to generate changes of less than 5% in the dose and 3 mm in position of the 90% isodose surface. In a study of the contribution of the tone and its matching collimator assembly to x-ray room Ieakage, it was noted that although the treatment tone had a negligible contribution, the upper annuli of the upper collimator assembly contributed as much as 80% of the leakage at 16 MeV for the 5cm tone. Electron cones, Electron dosimetry, Linear accelerators, Intraoperative radiotherapy.
INTRODUCTION
designed to deliver electron beams from 6 MeV to 18 MeV and to produce flat beams for field sizes as large as 15 cm in diameter. A number of clinical contraints were imposed upon the design. First, clinical experience suggested a tone length of 25-30 cm is needed to reach tumors deep within the abdomen. Second, each tone should be as light as possible. Third, the collimation system and tone should be easily sterilizable. Fourth, the collimation system should allow only minima1 leakage to the patient. Fifth, misalignment due to the laser docking system should minimally affect the dose distribution and radiation leakage. Sixth, the resulting dose distribution within the patient should be as uniform as possible inside the 90% isodose surface, and the geometrie coverage of the 90% isodose surface should be as large as possible. Seventh, the effect of the collimation system on x-ray leakage to the adjacent rooms should be minimal. This latter constraint is important in a treatment facility where the accelerator is being installed in an existing operating room, as there
This paper describes an approach and concepts in the design of a tone system for collimation in intraoperative electron therapy. A tone, which is a tube with a circular, rectangular, or other regularly shaped cross section through which the electrons pass in route to irradiating the target volume in the patient, is practica1 for this application. The wal1 of the tone shields the patient anatomy outside the tone from primary radiation. The homogeneity of dose to the patient’s target volume and the leakage dose outside the tone depend on the tone design and how it is interfaced to the therapy machine. A number of investigators ( 1,2, 3,9, 11) have designed or modified tone systems that attach directly to the head of the linear accelerator. In the present work, results are reported for a tone system designed for a dedicated intraoperative electron accelerator.* This system is unique in that it uses a laser docking system in which the patient tone is not in direct contact with the treatment head. The machine was
Reprint
Accepted for publication 9 August 1989. * Mevatron-ME, Siemens Medical Systems, CA.
requests
to: Kenneth R. Hogstrom, Ph.D. investigation was supported in part by a research contract with Siemens Medical Laboratories, Inc., and by a grant from the National Cancer Institute, CA-06294.
Acknowledgements-This
1223
Inc.. Concord.
1224
1. J. Radiation Oncology 0 Biology 0 Physics
may be structural limits to the amount of shielding that can be added. We discuss a number of design techniques, including determination of the thickness of circular trimmers (annuli) and the wal1 thickness of cones; determination of the geometry of annuli; and evaluation of internal blocking ring geometry. Two sets of circular cones with intemal diameters of 5 cm to 12 cm in 1-cm steps were specified, one with the lip of the tone perpendicular to the beam, and one with a 22.5” angle relative to the perpendicular plane. Figure 1 illustrates the location and geometry of the collimators. The upper collimator assembly consists of two annuli (upper and lower) fixed to the head of the machine. The clinical beam is defined by a third annulus (tone annulus) attached to the treatment tone and detached from the machine. lt is aligned with the treatment head via a laser docking system (Fig. 2), which employs four pairs of laser beams. When each pair (a line and a point) intersects to form four points every 90” along a 15.2 cm diameter circle, the treatment cone’s lateral position, distance from the treatment head, and angle are correct. Each laser spot is approximately 1 mm in diameter, and experience has shown agreement in alignment of the spots as smal1 as 1 mm are easily achievable. Therefore, alignment should be within 1 mm in the lateral dimension, 0.7 mm along centra1 axis, and 0.5” in angulation.
May 1990, Volume 18. Number 5 Laser Beams (8)
Fig. 2. Laser alignment system. Eight laser beams, four lines and four points, intersect on four points that lie on the circumference of a circle scribed on the alignment plate when the treatment tone is in proper alignment.
METHODS
AND
MATERIALS
Design theqv
Treatment
Head
UpperAnnulus
UPPer Collimating
Lower
1
Cone
Annulus
Assembly
Annulus
Fig. 1. Schematic of collimating system of the electron beam. The upper two annuli, forming the upper collimator assembly, are rigidly attached to the treatment head. There is a 15-cm air gap between it and the treatment tone.
In determining the size and shape of the electron collimator assembly and treatment tone, two types of electron leakage were considered: the transmission of electrons through the collimator material and the scattering of electrons outside the outer edge of the annuli. Collimutor material and thickness. The upper collimator annuli and the annulus on top of the treatment tone are brass. Brass was selected because of its high density, durability, machinability, and medium atomic number. It was felt that excessive bremsstrahlung production in a higher Z material would exacerbate room-to-room x-ray leakage. On the other hand, the low density of lower Z materials would require substantially thicker collimating annuli, perhaps leading to needless electron leakage from scatter off the annuli walls in the upper collimator assembly. Transmission measurements in brass (5) have shown that 12.7 mm is sufficiently thick to stop I8-MeV electrons. Both brass and acrylic (polymethylmethacrylate) were evaluated for the treatment tone. Acrylic offers the advantage of being transparent and has been frequently used by others for intraoperative electron therapy. Brass is durable, can be chrome plated, and flash sterilized. Its walls can be made thinner than acrylic, permitting better tumor coverage in tight-fitting anatomical situations. As a minimum, the tone wal1 thickness should be such that the thickness along a diagonal ray from the source is at least
Metallic cones for intraoperative electron therapy 0 K.
Electron Beam
the maximum range (R,,,) of a broad electron beam (in brass approximately 12.7 mm at IS MeV). The maximum diagonal thickness occurs for the largest field diameter and that portion of the tone closest to the source. When the source-to-surface distance is 100 cm, a 15-cm diameter, 30-cm long tone has a maximum diagonal wal1 thickness given by t,_,, = R,,,
* sin 8,,,
= 0.1 - R,,,.
2= fJx
$ (
‘T”.(Zi - Zi_1)2*Zi/6, i 81r
Air
(1)
Hence, a wal1 thickness of approximately 1.3 mm for brass and 8.8 mm for acrylic is required. The prototype treatment cones were only 25-cm long, and 1-mm brass wal1 thickness was selected. On the other hand, 8.8 mm acrylic is quite thick; therefore, 5 mm was selected for study based on the recommendations of McCullough and Biggs (10) of 3-5 mm. As noted in the results, greater wal1 thickness is required to control leakage through the side of the tone. Geometr)! o/‘collimating annuli. The size and location of the collimating annuli were determined using the clinical constraints described earlier and the theory of electron transport through air. As electrons pass through air they undergo multiple Coulomb scattering, which results in a lateral spreading of the electron beam. Lateral electron distributions in air can be calculated using Fermi-Eyges multiple scattering theory (7). In Figure 1, the first two annuli, the upper annulus and the lower annulus. form an upper collimator assembly that is attached to the treatment head. The third annulus is an integral part of the treatment tone itself. The lateral dimensions of the collimator annuli were selected to minimize weight and leakage. Figure 3 illustrates the principle of the collimator design (5). The penumbra generated by an upstream collimator must be shielded by a downstream collimator. For a collimator edge perpendicular to the centra1 axis, the outer edge of the downstream collimator should extend to at least 3.1 times cX outside the projected edge of the upstream collimator for the leakage due to electron scatter to be less than 2%; the inner edge should lie 1.3-1.6 times u, inside the projected edge to ensure uniformity of electron fluence within the aperture. cX is the root mean square value of the lateral spatial distribution of an electron pencil beam originating at the upstream collimator and arriving at the downstream collimator. It is derived from fundamental pencil beam equations (6. 7). resulting in
(2)
where Zi is the distance from the electron source to the top of the downstream collimator and Zi-1 is the distance from the source to the bottom of the upstream collimator. At sea level, the linear angular scattering power in air for the most probable incident energy, E,.o, equaling 6 MeV is approximated by (4)
1225
R. HOCSTROM et UI
I/
Fig. 3. Electron collimator design concept. The downstream annulus is designed to intercept the penumbra due to the upstream collimator (from the 2% profile intensity to the 95%’profile intensity).
= 2.3 - lom4 radian’/cm.
(3)
For calculation purposes. 6 MeV was selected because electrons scatter the most at the lowest energy. Our design theory strictly applies to rectangular fields. Rut when it is applied to circular fields, it results in an upper limit for leakage. Also for treatment facilities wel1 above sea level, where air scattering is less, this should again yield an upper limit. The upper and lower annuli diagrammed in Figure 1 are rigidly attached to the treatment head. The upper annulus is just below the collimation in the treatment head, and its outer edge extends outside the aperture in the treatment head so that no electrons can escape. The penumbra tast by the inner edge of the upper annulus is intercepted by the lower annulus. Similarly, the tone annulus is designed to intercept the penumbra tast by the lower annulus of the upper collimator assembly.
Misalignment
in plane perpendicular to centra1 axis.
Because the collimators were designed to be used in a “soft” docking system, translation errors in the alignment procedure must be allowed. It was felt that a 1-mm translation accuracy should be easily achievable; nonetheless, a conservative *5 mm misalignment criteria was incorporated into the final design. This means that the electron beam fluence, incident on the tone annulus should be uniform for an additiona10.5 cm in radius outside a given annulus aperture. For this reason the outer radius of the tone annulus was increased by 1 cm in the final design (not in the prototype) to prevent leakage, 0.5 cm to account for the broader diameter of the incident fluence,
1226
1. J. Radiation Oncology 0 Biology 0 Physics
and 0.5 cm to account for misalignment. This design feature adds additional weight to the tone, for example, 0.4 kg for the 9-cm diameter tone. Evaluation criteria. Al1 standard dosimetry quantities for electron beams are of interest: (a) surface dose, therapeutic depth, dose gradient, and bremsstrahlung for the depth-dose curve, (b) beam flatness, (c) the therapeutic isodose (90%) coverage; and (d) collimator leakage. For intraoperative therapy, not only is leakage measured in a phantom at the end of the tone important but so is leakage through the side of the tone because of healthy tissues that might be adjacent to the tone wall. Methods of measurement A variety of measurement conditions were used on this project because of the ongoing development of both the tone system and the accelerator. Data were taken at both Siemens Medical Laboratories, Inc. (SML) and M. D. Anderson Cancer Center (MDACC). Initial measurements were made at the SML factory on the intraoperative electron accelerator.* Five electron energies, 6, 9, 12, 15, and 18 MeV, were available. For the 18-MeV beam, a most probable energy of 17.4 MeV was the maximum achievable from the magnetrondriven accelerator. The prototype collimation system used 25 cm long brass and acrylic cones. Ionization measurements were made using a 0.14 cm3 cylindrical air ionization chamber (0.4 cm intemal diameter) and electrometer,+ which was scanned by an analog linear scanner. Off-axis stans were made both in air and in a water phantom. Ionization readings were normalized to 100% on centra1 axis. Film* measurements were made in a “solid-water” phantom. The film phantom had a cassette that accepted bare film and ensured that the film’s edge was precisely aligned (kO.1 mm) with the phantom surface. Film was exposed edge-on both beneath the tone in a plane containing centra1 axis and in a plane perpendicular to the tone wal1 and parallel to centra1 axis. Exposures were controlled so that the film optica1 density (O.D.) was within the linear portion of the dose-response curve (O.D. < 2). Film was developed by an automatie processor, and a nonirradiated film was used for determining background. The films were scanned at M. D. Anderson using a homemade scanning densitometer. Subsequent measurements were made using the 20 MeV beam and 7 cm tone assembly on a conventional radiotherapy accelerator* at M. D. Anderson to study the effect of plating a ring inside the tone. These data were taken with the M. D. Anderson data acquisition system,
* Mevatron-ME, Siemens Medical Systems, Creek, CA. + Capintec, Ramsey, NJ. * XTL-2 film, Kodak, Rochester, NY. p Mevatron KD, Siemens Medical Systems, Creek, CA.
Inc.,
Walnut
Inc.,
Walnut
May 1990, Volume 18, Number 5
which uses a 0.1 cm3 PTW air-ionization chamber and a PTW water phantom and scanner.** The charge was integrated by a computer-controlled electrometer.tt The entire system was controlled by a microprocessor*+ with software developed in house. The raw data was transferred into dose using a (3/4) radius displacement correction, a perturbation factor for the cylindrical chamber, and CE to convert the exposure in air to the dose in water. Following preliminary data on design of the internal ring, measurements were made at SML for the 7, 9, and 12-cm cones at energies of 6, 12, and 18 MeV, varying the position of the ring at 5, 10, and 15 cm above the bottom lip of the tone. The optimal position was determined to be 10 cm, and only samples of that data are reported here. Isoionization measurements were made using a 0.14 cm3 ionization chamber +in a Wellhofer water phantom and beam scanning system.@ The final measurements of tone leakage and misalignment studies were made at MDACC after installation of the accelerator. During installation it was decided to replace the 18 MeV beam with a 16 MeV beam so as to prolong the life and dependability of the magnetron. Isodose measurements were made by combining centra1 axis depthdose data, measured using the M. D. Anderson data acquisition system with off-axis profiles measured using film. RESULTS Collimator leakage In preliminary studies, collimator leakage was evaluated both (a) downstream and outside the edges of the annuli (assumed due to scatter and most significant at the lowest energy, 6 MeV) and (b) through the annuli and tone (assumed due to transmission and most significant at the highest energy, 18 MeV). Figure 4 is an in-air radial, relative ionization profile in a plane 1 cm below the lower annulus of the upper collimator assembly used with the 7-cm diameter tone. The low relative ionization under the lower annuli verifies that al1 primary electrons are stopped in the brass. Outside the outer edge of the lower annulus, the leakage radiation is less than 2%, and the beam intensity within the aperture is sufficiently uniform, verifying the proper dimensions of the upper collimator assembly as designed from scatter theory. In Figure 5, the in-air, radial, relative ionization profiles downstream of the upper collimator assembly at the position of the top of the treatment tone are plotted for the upper collimator assembly used with the 7-cm diameter tone. At both 6 MeV and 18 MeV, the annulus (prototype)
** Nuclear Associates, Carle Place, NY. ++Model 6 17, Keithley, Cleveland, OH. #* DEC LSI- 11/02. @ Frank Barker Associates, Pequannock,
NJ.
Metallic cones for intraoperative electron therapy 0 K. R.
-
18 MeV
---
8 MeV
HOGSTROM ef al.
1227
is removed in the final design as the aperture of the upper collimator assembly is increased to move the penumbra an additional 0.5 cm away from centra1 axis. The radial width of the tone annulus was increased 1 cm, 0.5 cm to account for the increase in the diameter of the upper aperture and 0.5 cm to allow for misalignment of the tone laterally. Cone leakage Previous tone designs have primarily used acrylic with walls as thin as 3 mm for 18 MeV electron beams (9). In
Lower Annulus of Upper Collimator Assembly
____-* 0 Radial Distance (cm)
-5
Fig. 4. Comparison of in-air radial ionization profiles for 6-MeV and 18-MeV beams measured 1 cm below the upper annulus. Hashed rectangles indicate the projection of the edges of the
lower annulus. on top of the 7-cm diameter
tone has been positioned to properly intercept the penumbra. Leakage outside the annulus is smal1 and believed mostly to be X rays. Inside the annulus, the beam has good uniformity at 18 MeV; however, at 6 MeV, the relative ionization is below 90% in a region extending from the edge to 0.5 cm inside the edge of the tone annulus. This region of nonuniformity 100
-Y \\
/--,’
: :
:
I I
z s f .s 9 t 3
-
18 MeV
---
8 MeV
Bc
this case, McCullough et al.‘s radial profile data in a water phantom below the tone show leakage in excess of 10% outside the penumbra at 18 MeV for the larger field sizes. We believe this leakage was probably due to transmission leakage through the side of the cones, which could result in substantial leakage at the outer surface of the cones. Fraass et al. (3) have shown that stainless steel wrapped around cones with 6-mm thick acrylic walls is necessary to eliminate leakage for a 20-MeV beam. However, neither of these studies measured leakage dose to tissue that may be adjacent to the outside walls of the collimator. To measure this effect, we exposed films in a solid-water phantom placed perpendicular to the outer surface of the prototype cones and parallel to the centra1 axis. Results of these measurements are shown for a 1-mm thick brass tone and a 5-mm thick acrylic tone in Figures 6 and 7, respectively. Both figures show electron leakage through the walls, which is greatest at the upper portion of the tone, where the electrons have the greatest angle of incidence. Note that the film lateral to the cones was exposed by delivering 10 times the number of monitor units used to expose the film in the treatment field. For the brass tone, the leakage is as great as 8% of the maximum dose on centra1 axis, and for the acrylic tone as large as 16%. The dose fall-off laterally was measured to be 50% in approximately 1.0 gm/cm2 of plastic phantom material. Based on these data, it was decided to increase the thickness of the walls by that equivalent amount. This would make the acrylic cones prohibitively thick (1 cm), but would only increase the brass cones by about another 1 mm.
50
I I I I
i,I
I\
:
\ Annulus on Top of : Treatment Cone !
-4& \ ;;
-5
0 Radial Distance km)
Fig. 5. Comparison of in-air radial ionization profiles for 6-MeV and 18-MeV beams measured at the loeation of the tone annulus. Hashed reetangles indicate the position of the edges of the tone annulus.
Dose distributions The two main features evaluated from the planar dose distribution were coverage of the 90% isodose contour (i.e., the treatment volume) and beam uniformity within that contour. Initial measurements clearly showed that both acrylic and brass cones had scatter off the tone walls resulting in rims of increased dose just inside the field edges at superhcial depths. This is illustrated in the radial, relative dose plot at a depth in water of 0.9 cm in Figure 8 for a 7-cm diameter tone at 18 MeV. Note the lower leakage outside the field of the brass tone (consistent with data of Figs. 6, 7) and that both cones exhibited a 20% hot spot along the rim. At 6 MeV such a comparison again yielded almost identical curves for acrylic and brass,
1228
1. J. Radiation Oncology ??Biology 0 Physics
May 1990, Volume 18, Numher 5
errors might cause significant changes in the dose distributions. For determination of ring geometry, a 20-MeV electron beam+ was used (an I8-MeV beam was not available at the time of testing) because the higher energy exhibited the greatest effect of tone scattering. An intermediate ring position, 10 cm above the bottom of the tone, was selected for the purpose of removing electrons streaming from the upper portion of the tone and decreasing the number of electrons striking the lower portion of the tone. The brass ring was 1.27 cm thick in the direction of the beam; its radial thickness was varied. Figure 9 illustrates isodose plots for use of no ring and rings with radial thicknesses of 0.25 cm and 0.125 cm. The wider ring reduced the hot spot of the rim, but sacrificed coverage of the 90% isodose line. It was originally decided by the radiotherapists that a 10% hot spot ( 110% isodose) would be clinically acceptable, so as not to lose coverage of the 90% isodose. A revised set of cones was then designed and constructed; these were 30 cm long and had walls 1.8-mm
-
7cm
I
-3/16*‘(6mm) Fig. 6. Demonstration of electron leakage through the wal1 of the brass tone for an 18-MeV beam. Note that the leakage radiation is greatest near the top of the tone.
but with a hot spot of only 5% and leakage of less than 1%. As the dose uniformity inside the tone was the same for the two types of material and the brass tone had significantly less leakage than the acrylic tone, we decided to use brass cones in the final design. Two methods of achieving beam uniformity have been previously studied: (a) adjustment of x-ray jaw width ( 1, 3, 11) to adjust the number of electrons striking the wal1 of tone, and (b) insertion of a smal1 collimation ring inside the tone (8). One problem with the first method is that the x-ray jaws form a square field for the circular cones so that beam uniformity on the major axis does not necessarily result in beam uniformity along the diagonals. For the machine we used, this is not a problem, as the circular annuli of the upper collimator assembly define the electron beam incident on the tone and can be adjusted accordingly. However, this method was abandoned and the second method employed because the tone is not rigidly attached to the treatment head and smal1 alignment
rm (5 YU En -2 Film
Fig. 7. Demonstration of electron leakage through the wal1 of an acrylic tone for an 18-MeV beam. Note that the leakage is greater than that of the brass tone in Figure 6.
Metallic cones
for
intraoperative
-
Bram
---
Acrylic
electron therapy ??K.
R. HOGSTROM efal.
1229
CR
I ___---
J I
1
-5
I 0
\ I
Na Ring
5
Radial Distance (cm)
Fig. 8. Comparison of radial-dose profiles in a water phantom at 0.9 cm depth for the acrylic and the brass cones with the 18MeV beam and 7-cm diameter tone. Note the 20% hot spot just inside the penumbra and the leakage just outside the penumbra.
thick for the top 20-cm length and 1.0 mm for the final 10 cm. The internal ring was placed 10 cm from the exit end of the tone. In light of the results shown in Figure 9, the ring radial thickness for each tone was adjusted so that a ray from the source grazing the inner edge of the lower surface of the ring would strike the end of the tone 6.7 cm above the bottom (e.g., 0.125-cm ring for a 7-cm tone). Isoionization curves were subsequently measured at 6, 12, and 18 MeV for the 5-, 9-, and 12-cm diameter cones at the SML factory. The resulting isoionization curve for the 12-MeV beam produced by the 9-cm tone is shown with the ring 10 cm above the bottom lip in Figure 10. Only isoionization contours less than 90% are shown when no ring was used. In summary, the ring has the effect of(a) reducing the hot spot, (b) decreasing slightly the coverage of the 90% isodose, and (c) broadening the lO%-30% isodose lines at depths less than d,,, . The reason for the latter is not fully understood.
Fig. 9. Comparison of isodose curves for varying ring radial thickness for a 20-MeV beam and 7-cm brass tone. The ring is located 10 cm above the bottom lip of the tone. The dashed curve indicates the position of the 90%’isodose in the absente of the ring.
Measurements were made at 6 MeV and the 7-cm tone misaligned by 1 mm, 2 mm, study errors in lateral translation. Based on surements in a plane containing centra1 axis, -
Rins ---NO
18 MeV with and 5 mm to isodose meano differente
Rins
Cone misalignment Because of the potential for tone misalignment when using a laser docking system, the dose distributions were measured with the cones intentionally misaligned. As stated earlier, we believe it is possible and practica1 to achieve alignment of the treatment tone with position errors less than 1 mm in lateral offsets and angles less than 0.5”.
10
I
I
I
-5
0
5
Radial Distance km)
Fig. 10. Isoionization curves for a 9-cm tone with a ring (solid lines) and without a ring (dashed lines) for 12 MeV electrons. Isodose curves above 90% are not indicated for the latter.
1230
1. J. Radiation Oncology 0 Biology 0 Physics
was detected within the accuracy of measurement (& l%, f 1 mm). This was not surprising because the cones were designed to permit a 5-mm error in lateral alignment. On the other hand, we were concerned about angulation errors because of the obvious sensitivity of the dose distributions to scattering off the inside walls of the collimator. Measurements were made at 6 MeV and 15 MeV with the 5-cm and 12-cm cones misaligned by O”, 0.5”, 1.O”, and 1.5”. Isodose measurements in a plane containing centra1 axis for the 5-cm tone using the 16 MeV beam are compared for the 0” and 1.O”misalignment in Figure ll. On the side farthest from the source (-), there is no significant change in the 90% isodose, and within the treatment volume (90% isodose for O”), an increase of O5% in dose is observed. On the side closest to the source, the 90% isodose line lies 0-3 mm inside the treatment volume, and within the treatment volume, a decrease of 0-5% in dose was observed. This type of effect was observed for the 12-cm tone and at 6 MeV. Studies at 0.5” showed the differences to be approximately 50% less and at 1.5” approximately 50% more. X-ray production by cones
The goal of the overall project was to develop an intraoperative radiotherapy suite within a conventional surgical suite using a machine that had only electron beam capability. Radiation shielding for the machine was necessary only for secondary x-ray production in the treat-
Radiel
Distance
(cm)
Aligned (Oo tilt)
Misaligned (le
May 1990, Volume 18, Number 5
ment head, the patient, and the collimating system. A set of measurements was made to determine the relative contribution of the collimating system to x-ray leakage. Measurements of x-ray leakage in the room directly under the machine were made for each of five energies for the electron beam directed downward (0”). These measurements were made at point A, located on centra1 axis 3.3 m below isocenter, and at point B, located at the same centra1 axis distance, but 3 m lateral to centra1 axis in the gun-target plane. Point A was selected as being directly under 20 cm of lead shielding in the floor, which shields the primary beam. Point B is selected to lie wel1 outside the primary beam and hence outside the lead shielding of the floor. The measurements were made for the 5-cm and 12-cm diameter cones at the highest energy achievable in the final installation, 16 MeV. Measurements were made (a) with the full collimation system in place, (b) with the tone removed, (c) with the tone and the lower annuli of the upper collimator assembly removed, and (d) with both the tone and the upper collimator assembly removed. In al1 cases, the electron beam was absorbed by a 25 cm deep water phantom at the 100 cm SSD. The results of these measurements are listed in Table 1. Removal of the tone resulted in only a modest decrease in the leakage radiation, being more significant for the higher energy, larger tone, and point B outside the beam. Qualitatively, this was expected, as x-ray production increases with energy, the larger tone has more surface area, and there is little attenuation of radiation produced in the tone by the floor at point B. The magnitude of leakage produced in the tone is insignificant from a safety perspective. Removal of the lower annulus of the upper collimator assembly has an even less significant effect. On the other hand, a significant reduction in leakage was observed with the removal of the upper annulus, indicating it was a major source of leakage. As expected qualitatively, the reduction in leakage was most significant for the higher energy, smaller tone, and point B outside the beam. In this case, the smaller tone has the largest upper annulus in the upper collimator assembly, hence producing the most X rays. At 16 MeV for the 5-cm tone and at point B, a decrease of 83% is observed, which is significant from a safety perspective. The upper annulus appears to be the major contributor to leakage radiation; therefore, efforts should be directed toward replacing al1or part of the upper aperture with a lower Z material such as aluminum (p = 2.7), beryllium (p = 1.8), or encased graphite (p = 2.3).
tilt)
SUMMARY
Fig. ll. Comparison of isodose curves for the 5-cm diameter tone at 16 MeV, for the tone properly aligned with those for the tone angulated 1’. The dashed curve indicates the position of the 90% isodose when the tone is aligned.
It has been demonstrated that the lateral dimensions of electron collimators (annuli) can be determined using multiple scattering theory at the lowest electron energy and that the thickness can be determined from the maximum range, R,,, measured for electrons in the collimator materials at the highest energy. On this basis, the
1231
Metallic cones for intraoperative electron therapy 0 K. R. HOGSTROM etal.
Table 1. Influence of collimation Collimation present*
Location 12-cm tone Point A
Point B
5-cm tone Point A
Point B
6 MeV
on x-ray leakage (al1 data in mR - h-‘) 9 MeV
12 MeV
15 MeV
16 MeV
1 2 3 4
1.9 1.9 1.9 1.8
5.5 5.2 5.2 4.8
11.0 9.4 9.6 8.4
18.0 17.5 17.0 14.5
21.0 18.0 18.0 15.0
1 2 3 4
4.0 3.8 3.6 2.0
11.0
10.5 10.0 4.2
21.0 20.0 19.0 6.0
42.0 34.0 33.0 9.4
42.0 37.0 35.0 9.4
1 2 3 4
2.5 2.4 2.3 1.8
6.4 6.2 6.2 4.6
12.0 12.0 12.0 8.2
23.0 21.0 21.0 15.0
24.0 23.0 23.0 15.0
1 2 3 4
5.4 5.4 5.2 2.0
15.0 15.0 15.0 4.4
30.0 28.0 28.0 6.0
50.0 49.0 48.0 9.4
56.0 54.0 54.0 9.4
* 1 = upper aperture and lower aperture of upper collimator assembly and tone; 2 = upper aperture and lower aperture of upper collimator assembly; 3 = upper aperture of upper collimator assembly; 4 = upper collimator assembly and tone removed.
collimation
system
for a set of circular
brass cones has
been designed for a radiotherapy machine* specially designed for intraoperative electron therapy. Note, that careful attention must be paid to the leakage through the side of a treatment tone, primarily due to electron scatter. At 18 MeV, leakage through 5-mm acrylic cones was as much as 16% and through brass cones with 1-mm radial thickness as much as 8%. In both cases, the radial thickness must be approximately doubled to reduce the electron leakage by half. Hot spots in the isodose distributions for the radial planes, as great as 120%, were observed due to scatter off the inside wal1 of the tone. In an effort to remove this rim of increased dose, an internal ring was placed inside the tone 10 cm upstream of the bottom lip of the tone. Results showed that a radial width, whose inside edge projects 6.7 cm from the bottom lip of the tone, reduces the hot spot to approximately 110%. By increasing the radial thickness of the ring, the hot spot can be further reduced; however, each reduction in the hot spot is accompanied by a corresponding decrease in the coverage of the 90% isodose surface. Therefore, any evaluation of clinical advantages of this ring must weigh more uniform dose against poorer coverage of the treatment volume at depth. Because the treatment tone is docked using a laser alignment system, the tone may be mispositioned. A study of these errors showed that a 5-mm translation error had no significant effect on the relative dose distribution. Translation errors greater than 2 mm are not expected. Data show that a 1o alignment error increases the dose by as much as 5% on the side of the tone farthest from
the source and decreases the dose by as much as 5% on the opposite side. Rotation errors greater than 1’ are not expected, so this is not expected to be a clinical problem. A study of radiation leakage to the room below revealed that the electron collimation system was a major source of x-ray leakage. The electron tone assembly and the lower annuli of the upper collimator assembly contributed negligibly to the leakage. On the other hand, the upper annulus of the upper collimator assembly contributed as much as 80% of the leakage for the smal1 field sizes at the highest energy, 16 MeV. In other words, that annulus is intercepting a significant portion of the beam and is serving as a secondary x-ray source. It should be possible to reduce this leakage by replacing al1 or part of the upper annuli, made of brass, with one of a lower Z material. During the course of these studies, a computer program that designs electron treatment cones was developed. This software was used to design a circular set of cones with diameters from 5 to 12 cm in 1-cm steps and the following features: 30-cm length, brass composition, 1.78 mm wal1 thickness the upper 20 cm, 1 mm the lower 10 cm, allowance for a 0.5-cm misalignment, and no internal ring. The radiotherapist’s decision to remove the ring was based on the philosophy that the benefit of the increased coverage of the 90% isodose line was worth the smal1 volume of increased dose. The resulting cones weigh 1.9 kg for the 7 cm diameter, 2.4 kg for a 9 cm diameter, and 3.0 kg for the 12-cm diameter tone. The collimation system can produce good electron dose distributions reliably and safely during intraoperative radiotherapy and appears suitable for clinical use.
232
1. J. Radiation Oncology 0 Biology 0 Physics
May 1990. Volume 18, Number 5
REFERENCES 1. Bagne, F. R.; Samsami, N.; Dobelbower, R. R. Radiation contamination and leakage assessment of intraoperative electron applicators. Med. Phys. 15530-537: 1988. 2. Biggs, P. J.; Epp, E. R.; Ling. C. C.: Novack, D. H.; Michaels. H. B. Dosimetry, field shaping and other considerations for intra-operative electron therapy. Int. J. Radiat. Oncol. Biol. Phys. 7:875-884; 198 1. 3. Fraass, B. A.; Miller, R. W.; Kinsella, T. J.; Sindelar, W. F.; Harrington, F. S.; Yeakel, K.: Van De Geihn, J.; Glatstein, E. Intraoperative radiation therapy at the National Cancer Institute: Technical innovations and dosimetry. Int. J. Radiat. Oncol. Biol. Phys. 1 1: 1299- 13 11; 1985. 4. Hogstrom, K. R.; Kurup, R. J.; Shiu, A. S.; Starkschall, G. A two-dimensional pencil-beam algorithm for calculation of arc electron dose distributions. Phys. Med. Biol. 34:3 I5341; 1989. 5. Hogstrom, K. R.; Meyer, J. A.; Melson, R. Variable electron collimator for the Mevatron 77: design and dosimetry. In: Proceedings of the 1985 Mevatron Users Conference. Iselin, N.J.: Siemens Medical Systems, Inc.; 1986:25 1-276.
6. Hogstrom, K. R.; Mills, M. D.; Almond. P. R. Electron beam dose calculations. Phys. Med. Biel. 26:445-459; 198 1. 7. Huizenga, H.: Storchi, P. R. M. The in-air scattering of clinical electron beams as produced by accelerators with scanning beams and diaphragm collimators. Phys. Med. Biol. 32:101 1-1029; 1987. 8. Kao, M.; Lanzl, L.: Rozenfeld, M.: Pagnamenta. A. Dose uniformity of the intraoperative radiation therapy tone (Abstract). Med. Phys. 13:605: 1986. 9. McCullough, E. C.; Anderson, J. A. The dosimetric properties of an applicator system for intraoperative electronbeam therapy utilizing a Clinac- 18 accelerator. Med. Phys. 9:261-268; 1982. 10. McCullough, E. C.; Biggs, P. Physical aspects of intraoperative electron beam radiation. Curr. Probl. Cancer 7:2430: 1983. 11 Ochran, T. G.; Schabinger, P. R. Clinical application and dosimetric evaluation of a state-of-the-art intraoperative tone system. In: Proceedings of the 1987 Therapy Users Conference. Iselin, NJ: Siemens Medical Systems, Inc.: 1987: 2 15-240.
