Phys. Med. Biol. 42 (1997) 15051514.

Printed in the UK

PII: S0031-9155(97)81132-3

Practical approach to electron beam dosimetry at extended SSD

Joanna Cygler, X Allen Li, George X Ding and Edward Lawrence
Department of Medical Physics, Ottawa Regional Cancer Centre, 190 Melrose Avenue, Ottawa, Ontario K1Y 4K7, Canada Received 20 January 1997 Abstract. This paper presents a practical approach to clinical dosimetry for electron beams at extended source-to-surface distance (SSD). Characteristics of electron beams from a Siemens MD-2 accelerator are presented at a nominal SSD of 100 cm and an extended SSD of 115 cm. Relative output factors are measured at both 100 and 115 cm SSDs for a range of square eld sizes for each beam energy. The change of output with SSD does not follow the inverse square law, if the nominal SSD is used. This deviation is larger for lower beam energy and smaller eld sizes. An effective SSD, SSDeff , has been determined at dmax for each beam energy as a function of eld size. A comprehensive approach to treatment time calculations is proposed. Civic Division,

1. Introduction The characteristics of clinical electron beams depend on the primary beams parameters as well as on the scattering materials present in the beam. Electron interactions with the accelerator head components and the eld dening apertures affect the dose distribution in the phantom not only at a standard source-to-surface distance (SSD) but also at extended treatment distances. The fraction of electrons scattered from the applicator walls and reaching the phantom has an inuence on the percentage depthdose curve (%DD), eld atness, penumbra and relative output factors (ROF). These beam characteristics change when the SSD increases. It has been noted before that corrections to dose rate at extended SSD do not follow the inverse square law (ISL) if the nominal value of SSD (usually 100 cm) is used (Khan et al 1978). This is due to the fact that the nominal SSD is often dened as the distance from the accelerator exit window to the phantom surface, while the apparent source is in fact positioned not at the window but at various distances downstream from the window, depending on the amount of scattering material present in the beam (ICRU 1984). Several approaches have been proposed to describe the electron source in a clinical accelerator: effective extended source (ICRU 1984), virtual source (Khan et al 1991) and effective source (Khan et al 1978). They all have their merit for various applications but also have their limitations. For the purpose of ISL corrections to the dose rate in clinical situations, the effective source approach proposed by Khan et al (1978) is the most practical one, since it properly predicts the distance dependence of the output of clinical beams. The effective SSD is the distance from the phantom surface to the effective source. The effective SSD is known to depend on beam energy and eld size (Khan et al 1978, Sweeney et al 1981, Jamshidi et al
0031-9155/97/081505+10$19.50 c 1997 IOP Publishing Ltd

1505

1506

J Cygler et al

1986, Roback et al 1995). This complicates the treatment time calculations at extended SSDs, since a large number of clinical data are needed. In our clinic we have recently commissioned a new Siemens MD-2 accelerator. This was an occasion to review our electron beam dosimetry and dene a more comprehensive approach to everyday clinical problems. In this paper we discuss the effects of extended SSD on electron beams characteristics as well as our approach to treatment time calculations. 2. Experiment A Siemens MD-2 accelerator produces electron beams in the energy range of 515 MeV. The unit has various applicators with nominal source to applicator-end-distance of 95 cm. This introduces a 5 cm air gap between the applicator end and the nominal SSD (100 cm) plane. The detailed description of this applicator has been given elsewhere (Ebert and Hoban 1995). A variety of eld-shaping cerrobend cut-outs were designed and inserted in the applicators to dene irregular elds. An RFA 300 dosimetry system (Therados) with a p-type silicon diode was used to measure lateral beam proles and depthdose curves in water for all eld sizes dened by open applicators and cut-outs. The output factors for each applicator and cut-out were measured at dmax for each eld size using an RK cylindrical chamber (0.12 cc) supplied by Scanditronix. The relative output factors for applicators (ROFapp ) were dened in reference to open 10 10 cm2 applicator. The relative output factors for cerrobend cut-outs (ROFcut ) were dened in reference to open applicators. Measurements were performed at various SSDs ranging from 95 to 125 cm. The measurement uncertainty was 0.3%. 3. Results 3.1. Depthdose curves The parameters of the %DD curves for electron beams for 100 cm SSD and open 10 10 cm2 applicator are presented as a function of the nominal beam energy in table 1. For a given eld size these parameters depend strongly on the beam energy. For a given energy, the depthdose curve depends on the eld size. For smaller eld sizes the maximum of the %DD curve shifts towards the surface of the phantom, as shown in table 2. This shift is more pronounced for higher energies.
Table 1. Siemens MD2 electron beam parameters, measured in water with silicon p-type diodes. 10 10 cm2 applicator, SSD = 100 cm. Nominal energy (MeV) 6 9 11 13 dmax (cm) 1.45 2.10 2.60 2.90 R85 (cm) 1.95 2.94 3.63 4.20 R50 (cm) 2.40 3.59 4.43 5.13 R20 (cm) 2.76 4.11 5.05 5.86 Rp (cm) 2.94 4.33 5.37 6.21

For eld sizes greater than 5 5 cm2 , it is found that as the SSD increases the percentage surface dose and dose in the build-up region decrease, but the %DD curves beyond the depth of maximum dose are not practically affected. The effects of extended SSD become more observable for higher electron-beam energies. It is found that the variation in the %DD

Electron beam dosimetry at extended SSD

1507

Table 2. Dependence of dmax (cm) on the cut-out size. 10 10 cm2 applicator, SSD = 100 cm. Square cut-out (cm2 ) Energy (MeV) 6 9 11 13 10 10 open 1.45 2.10 2.60 2.90 88 1.45 2.10 2.60 2.90 77 1.45 2.10 2.60 2.90 66 1.40 2.10 2.60 2.90 55 1.40 2.10 2.45 2.70 44 1.40 2.00 2.25 2.50 33 1.25 1.70 1.95 2.00 22 1.00 1.10 1.20 1.30

curves is up to 6% as SSD increases from 100 to 115 cm for a 13 MeV electron beam and 10 10 cm2 open applicator. The extended SSD effects observed at present for eld sizes greater than 5 5 cm2 agree generally with those reported previously (Saw et al 1994, 1995) for a similar machine. For cut-out sizes less than 5 5 cm2 , the extended SSD affects the entire %DD curves. Figure 1 shows the extended SSD effects on the %DD curves for the 6 and 13 MeV beams with the 2 2 cm2 cut-out in the 10 10 cm2 applicator. It is seen from gure 1 that the change in the %DD curves is up to 10% as SSD increases from 100 to 115 cm for the 6 MeV electron beam with the 2 2 cm2 cut-out. The data presented here for small eld sizes for the Siemens MD2 linac show comparable effects to those observed for a Varian Clinac-2100C unit (Das et al 1995). It should be noted that the biggest change due to extended SSD for a 10 10 cm2 eld occurs at the surface. Unlike the conclusions reported by the previous studies (Saw et al 1994, 1995, Das et al 1995), in the present work the percentage surface dose does not always decrease as the SSD increases. The percentage surface dose increases in these small-eld cases for which dmax shifts towards the surface. The percentage surface dose decreases when dmax shifts away from the surface or does not change. Note that the AAPM TG-25 (Khan et al 1991) suggests using a 1/r 2 divergence factor (TG25, equation (29)) to correct %DD at extended SSD. We nd that this divergence factor is too small to modify the %DD curves for all the eld sizes for our machine. For comparison, gure 1 includes the %DD curve for the 6 MeV, 2 2 cm2 cut-out, at SSD = 115 cm, corrected using the divergence factor. 3.2. Lateral proles The extended SSD effects on proles are represented by a loss of eld atness and an enlargement of the penumbra. Figures 2 and 3 show beam penumbra and atness against electron beam energy for different cone and cut-out sizes at SSD of 100 and 115 cm. The penumbra is typically dened as the distance between 20 and 80% of maximum intensity seen in the prole at the depth of maximum dose (ICRU 1984). The atness is specied as the percentage difference between maximum and minimum intensity seen in 80% of the eld width on the beam prole at depth of maximum dose, where the eld width is dened as width of 50% intensity (ICRU 1984). It is seen from gure 2 that the electron eld penumbra increases at extended SSD for all cone and cut-out sizes studied. This effect becomes more severe for lower electron energies. This is because electron scattering power is much higher for lower energies (ICRU 1984, Li and Rogers 1995). The loss of electron beam atness, which is represented by a rounded shape prole, depends slightly on cone or cut-out size and beam energy as seen from gure 3. This loss may be explained by the

1508

J Cygler et al

Figure 1. Percent depthdose curves for a 2 2 cm2 cut-out in a 10 10 cm2 cone at 100 and 115 cm SSD for (a ) 6 MeV and (b ) 13 MeV electron beams.

fact that the electrons scattered at large angles from the sides of the cone or the cut-out contribute to the dose at the edges of the eld at standard SSD. These electrons are lost from the beam at extended SSD and thus the beams are less at. 3.3. Relative output factors For clinical purposes relative output factors are needed for all applicators, cut-outs and SSDs. The applicator factors, ROFapp , dened as the ratio of the chamber reading at dmax for a given applicator at an SSD of 100 cm to the chamber reading at dmax for the 10 10 cm2 applicator at an SSD of 100 cm, are shown in table 3 for various beam energies. The ROFcut values for the cut-outs were normalized to the detector reading for the given open applicator

Electron beam dosimetry at extended SSD

1509

Figure 2. Electron eld penumbra versus electron energy for 2 2, 3 3 and 5 5 cm2 cut-outs in a 10 10 cm2 applicator and for 10 10 and 15 15 cm2 open applicators at 100 and 115 cm SSD. The penumbra is dened as the distance between 20 and 80% of maximum intensity seen in the prole at the depth of maximum dose.

Table 3. Siemens MD2 applicator factors, ROFapp (measured with RK chamber, normalized to 10 10 cm2 applicator at the standard SSD of 100 cm). Measurement uncertainty was 0.3%. Energy (MeV) 6 9 11 13 Circular 5 cm 0.834 0.924 0.938 0.954 10 10 cm2 1.000 1.000 1.000 1.000 15 15 cm2 1.013 0.995 0.993 0.993 20 20 cm2 1.022 0.974 0.972 0.965

at a given SSD. The data for cut-outs in the 10 10 cm2 applicator for SSDs of 100 and 115 cm are presented in gures 4(a ) and (b ) for 6 and 13 MeV respectively. The ROFcut values decrease for smaller cut-outs. The decrease is more pronounced for lower-energy beams. For the same beam energy, the decrease of ROF values, with the eld size is more pronounced for larger SSD. This is due to the loss from the beam of electrons scattered at larger angles. This loss is larger for longer SSDs. Similar data have been collected and similar trends have been observed for other applicators.

3.4. Effective source position and inverse square law To check the application of the inverse square law, and to nd the effective source position, measurements were performed at various SSD values ranging from 95 to 125 cm. The method we use to nd the effective SSD was proposed by Khan et al (1984). If Q0 is the ionization charge reading at dmax at the standard nominal SSD (100 cm) and Qg is the

1510

J Cygler et al

Figure 3. Electron eld atness versus electron energy for 2 2, 3 3 and 5 5 cm2 cut-outs in a 10 10 cm2 cone and for 10 10 and 15 15 cm2 open cones at 100 and 115 cm SSD. The atness is specied as percentage difference between maximum and minimum intensity seen in 80% of the eld width on the beam prole at depth of maximum dose, where the eld width is dened as the width of 50% intensity.

reading at dmax with the extra gap, then Q0 Qg

1/2

g +1 SSDeff + dmax

(1)

where the gap, g , is the difference between the extended and nominal SSD. From the slope of the straight line described by equation (1), one can calculate the effective SSD by 1 (2) dmax . SSDeff = slope Figure 5 presents (Q0 /Qg )1/2 as a function of the air gap introduced between the nominal and extended SSD for a 10 10 cm2 open applicator. The measured data follow a straight line for each electron energy; however, the slopes of the lines are different leading to different effective SSDs for different electron beams dened by the 10 10 cm2 open applicator. Figure 6 plots the effective SSD as function of eld size. It is found that, for 6 MeV electrons, the value of SSDeff is 82 cm for the open 10 10 cm2 applicator and only 31 cm for the 3 3 cut-out. 4. Discussion Electron interactions with the accelerator head components and the eld-dening apertures affect the dose distribution in the phantom. The effect of extended SSD on %DD curves depends on the eld size. For large eld sizes, there are only small effects observed in the build-up region. The percentage dose in the build-up region is lower at extended than at the standard SSD. However, these differences which are at most 2% are not clinically

Electron beam dosimetry at extended SSD

1511

Figure 4. Cut-out factors (ROFcut ) for (a ) 6 MeV and (b ) 13 MeV for cut-outs in a 10 10 cm2 applicator at 100 cm SSD. The measurement uncertainty was 0.3%.

signicant. More pronounced changes in %DD curves at extended SSD are observed for small eld sizes (gure 1). The curve is modied over the entire depth range. The change in %DD curve is specially apparent for lower energies. The observed effect suggests that beams dened by small cut-outs are more monodirectional and monoenergetic at larger SSDs than at shorter SSDs. The penumbra width increases dramatically at extended SSD, specially for lower beam energies (gure 2). This should be taken into account during treatment planning either by addition of an appropriately larger margin around the tumour volume or by putting the nal eld-dening cut-out directly on the patients skin. At the standard SSD the beam atness for a given eld size is practically independent of energy (gure 3). For a given energy, the atness is worse for smaller eld sizes.

1512

J Cygler et al

Figure 5. Determination of effective SSD (equation (1)) for an open 10 10 cm2 applicator. The estimated uncertainty was 0.5%.

Figure 6. Effective SSD as a function of cut-out size for beam energies 613 MeV. The measurement uncertainty was 0.5%.

Based on graphs similar to gure 5, the effective SSD is found for all beam energies. The data are presented in gure 6. For a given energy, the effective SSD depends strongly on the collimation opening. For larger collimation openings the effective source position appears to deviate less from the nominal value of 100 cm. Small cut-outs introduce more electrons scattered at larger angles. Therefore the angular distribution of electrons immediately below

Electron beam dosimetry at extended SSD

1513

Figure 7. Applicator gap factors, dened as the ratio of the chamber readings for a given open applicator at dmax at an SSD of 115 cm to the chamber reading for this applicator at SSD = 100 cm.

the cut-out is wider. This moves the effective source position closer to the phantom surface as compared to the larger eld size. For a given collimation opening, the effective SSD depends on the energy of the beam. The lower the beam energy, the shorter the value of effective SSD for all eld sizes below 15 15 cm2 . For eld sizes 15 15 cm2 and larger, the effective SSD is practically independent of energy. This behaviour can also be explained in terms of the angular distribution of the beam. The lower the energy, the wider the angular distribution of the beam and the shorter the effective SSD. For large eld sizes (>15 15 cm2 ) the electrons scattered at large angles from the eld dening apertures do not reach the central axis at dmax . This causes the angular distribution of electrons on the central axis to be narrower and the effective source of the beam appears to be right at the exit window of the accelerator for all beam energies. The error committed if the nominal SSD is used in the inverse square law corrections of the dose rate at extended SSD is larger for smaller energies and eld sizes. To reduce the chance of errors in treatment time calculations, we have decided to limit the number of dosimetric data for standard clinical use. Only one value of extended SSD is allowed, equal to 115 cm. Since our machines applicators are at 95 cm from the nominal source position, 115 cm SSD gives a 20 cm air gap between the end of the applicator and the patient surface. This is sufcient for all clinical situations. The treatment time is calculated based on the following formula: #MU = D RDR ROFapp GF ROFcut

where D is the dose per fraction, RDR is the reference dose rate for 10 10 cm open applicator at dmax and standard SSD of 100 cm, ROFapp is the applicator factor, GF is the applicator gap factor (GF = 1 at an SSD of 100 cm), and ROFcut is the cut-out factor for a given applicator at a given SSD. The applicator gap factors, GF, dened as the ratio of the chamber readings for a given open applicator at dmax at an SSD of 115 cm to the chamber

1514

J Cygler et al

reading for this applicator at an SSD of 100 cm are presented in gure 7. The cut-out factors, ROFcut , are presented in gure 4 for 6 and 13 MeV. It is seen from gure 4 that the cut-out factors are smaller at 115 cm than at 100 cm SSD. This effect is much more pronounced for 6 than for 13 MeV. It can be explained by a larger fraction of electrons with wider angular distribution in the 6 than in the 13 MeV beam. For a given cut-out size these electrons are lost from the beam when the air gap between the cut-out and the phantom surface is larger, leading to smaller ROFcut at extended SSD. 5. Conclusions Electron interactions with the accelerator head components and additional beam-dening cut-outs are complex and result in different values of effective SSD, SSDeff , for different energies and eld sizes. The inverse square law does apply to dose rate corrections at the range of studied distances, provided the proper value of SSDeff is used. Since this would require a rather large set of data, it is practical to limit the range of treatment distances for clinical use. It should be noted that the percentage depthdose is modied over the entire range of the beam at extended SSD for small eld sizes. For the given depth, penumbra increases with the increase of the treatment distance. All considerations about penumbra sizes should be passed to the physician, to make sure that adequate tumour coverage is achieved. References
Das L J, McGee K P and Cheng C W 1995 Electron-beam characteristics at extended treatment distances Med. Phys. 22 166774 Ebert M A and Hoban P W 1995 A model for electron beam applicator scatter Med. Phys. 22 141929 ICRU 1984 Radiation dosimetry: electron beams with energies between 1 and 50 MeV ICRU Report 35 Jamshidi A, Kuchnir F T and Reft C S 1986 Determination of the source position for the electron beams from a high energy linear accelerator Med. Phys. 13 9428 Khan F M, Doppke K P, Hogstrom K R, Kutchner G J, Nath R, Prasad S C, Purdy J A, Rozenfeld M and Werner B L 1991 Clinical electron beam dosimetry: report of AAPM Radiation Therapy Committee, Task Group 25 Med. Phys. 18 73109 Khan F M, Sewchand W W and Levitt S H 1978 Effect of air space on depth dose in electron beam therapy Radiology 126 24951 Li X A and Rogers D W O 1995 Electron mass scattering powers: Monte Carlo and analytical calculations Med. Phys. 22 53141 Roback D M, Khan F M, Gibbons J P and Sethi A 1995 Effective SSD for electron beams as a function of energy and beam collimation Med. Phys. 22 20935 Saw C S, Ayyangar K M, Pawlicki T and Korb L J 1995 Dose distribution considerations of medium energy electron beams at extended source-to-surface distance Int. J. Radiat. Oncol. Biol. Phys. 32 15964 Saw C B, Pawlicki T, Korb L J and Wu A 1994 Effects of extended SSD on electron-beam depthdose curves Med. Dosim. 19 7781 Sweeney L E, Gur D and Bukovitz A G 1981 Scatter component and its effect on virtual source and electron beam quality Int. J. Radiat. Oncol. Biol. Phys. 7 96771