Radiotherapy and Oncology.

I I (I 988) 395-404
Elsevier 395

RTO 463

alculation of electron beam de s

I.A.D. Bruinvis and W.A.F. Mathol

Key words: Electron therapy: Treatment planning: Irregular field

Summary

A previously presented method to calculate depth-dose curves and output factors for arbitrarily shaped
electron beams [6] is evaluated. The method employs a Gaussian pencil model for direct incident and appli-
cator scattered electrons; the parameter values are derived from measured central axis depth-dose distribu-
tions. In addition, an empirical model is used to compute the dose due to electrons scattered by fieid-defining
frames. In this way, the properties of the clinical electron beams are taken into account. In this paper,
calculations and measurements for electron beams with energies between 6 and 20 MeV. treatment field
dimensions between 3 and I4 cm, and various applicator sizes are compared. The results demonstrate the
importance of irregular field dose calculations and the scope of the present method. Agreement better than
3% in dose and 0.2 cm in depth is achieved. For electron beams without applicators, the calculations show
the same accuracy. Another method in electron treatment planning that derives values for the radial width
parameter of the pencil beam from measured broad beam profiles is also investigated. This method gives
good results for dose calculations in beams without applicator scatter. It should be used with care, however.
for beams that contain such a scatter component. When electrons scattered by the applicator walls and
field-defining frames are neglected, differences between measured and calculated dose up to 8% are found.

Introduction on the shape and size of the treatment field. Re-

ducing the field size can have a significant influence
In radiation therapy with high-energy electron on the shape of the depth-dose curve and on the
beams, the absorbed dose distribution produced in maximum dose value [l&22,23,32]. This also holds
tissue depends primarily on the electron energy and for narrow parts of an arbitrarily shaped treatment
field. These effects are stronger when the beam con-
tains electrons scattered by collimating devices. in
Address for mrrespondeme: I.A.D. Bruinvis, Radiotherapy De-
partment, Antoni van Leeuwenhoekhuis. Plesmanlaan 121, 1066 particular, applicator walls and field-defining
CX Amsterdam, The Netherlands. frames [ 18,20,34].

0167-8140/88/$03.50 0 1988 Elsevier Science Publishers B.V. (Biomedical Division)

396

Because irregularly shaped electron fields are [ll] type of calculation [12,19,31,37,38]. Some au-
used in many centres [2,10,16,27,28,38], a method thors do not consider the possible change in the
to calculate the dose distribution in three dimen- depth of maximum dose with a smaller field size,
sions as a function of field contour and electron however, which makes their methods incomplete.
energy is needed. Dose measurements for each in- A major disadvantage of the Clarkson-type
dividual field would constitute an unacceptable methods is the amount of dosimetric data to be
workload. Although modern systems of electron measured and stored in the computer.
beam collimation avoid applicator-wall scatter as Mills et al. [24] were the first to employ a Gaus-
much as possible [18]. many centres still use electron sian pencil beam model to predict output factors
beams that contain a certain amount of wdi scat- for square and rectangular fields defined by the ac-
tered electrons. The calculation method should celerator collimating blocks; they indicated how
therefore be capable of dealing with such a beam their method could also handle irregularly shaped
component. fields. Loyd and Rosen [21] applied an extended
Before discussing the literature on dose calcula- pencil beam model to arbitrary field shapes. Their
tions for arbitrarily shaped electron beams, we will method appears to be unnecessarily complicated
introduce some concepts used in this paper. The and time-consuming, considering the obtained re-
ICRU [18] defines the reference axis (or beam axis) sults.
as the line passing through the centre of the effec- In a previous paper [6], we presented our ap-
tive radiation source or the virtual point source and proach to the calculation of dose distributions for
the centre of the radiation field (for circular and arbitrarily shaped fields of clinical electron beams,
rectangular fields). In order to generalise this defi- also using a Gaussian pencil beam model. The vari-
nition for arbitrary field shapes, we define a cieptk ation in isodose distributions and in output factors
axis as a line parallel to the beam axis passing due to shielding within a 10 cm x 10 cm applicator,
through a specific point in the radiation field. A plot was computed with good results [6,7]. The values
of the absorbed dose values along this axis, ex- for the dose distribution parameters of the pencil
pressed as a percentage of the maximum dose on beam at each depth were derived from measured
this axis, versus depth is called a cieptk-dose CWPP. beam profiles and depth-dose curves. Separate cal-
The (local) output fictor’ is defined as the ratio of culations for electrons scattered by the field shaping
the maximum dose on the chosen depth axis in the device were performed. In this way, we took the
field of interest to that on the central axis of a ref- specific properties of the collimated electron beams
erence field [24]. For this reference field, we take the into account.
largest field within the same applicator as for the In this paper, the extension of our method to a
field of interest. The maximum central axis dose complete range of electron energies and applicator
values in the applicator fields are related to that in sizes is presented and attention is Focussed on the
the standard 10 cm x 10 cm field by measurements. computation of depth-dose curves and output fac-
Computing depth-dose curves and output factors tors. The applicability for electron beams with var-
as a function of field size from a limited number of ious amounts of applicator-scattered electrons is
experimental data has been the subject of many studied and two methods to derive values of the
studies. For rectangular, square and circular fields, pencil beam radial spread parameter are compared.
several techniques of interpolation between mea- Percentage depth-dose curves and output factors
sured dose values have been proposed and experi- calculated for various field sizes and shapes are
mentally verified (1,4,17,23-26,33,36]. An applica- compared with measured data. It is shown that sig-
tion of an equivalent square-field method to irregu- nificant errors can occur when the influence of elec-
lar fiehi shapes [28] has not proven to be adequate. trons scattered by applicator walls and field defin-
Better results have been obtained with a Clarkson ing frames is not taken into account.
 397

An extensive description of the calcuiation model

and the methods to derive parameter values is given
The electron beams investigated in this study were elsewhere [9]. Here. only an overview is given.
produced by means of a single scattering foil, solid-
wall applicators and field defining metal frames of
high atomic number. In some experiments, no ap-
plicator was used. The frame was in contact with A Gaussian pencil beam model [3,15,17] was used
the phantom surface. The major contribution to the for the dose distribution resulting from the elec-
dose is given by the electrons that arrive directly trons coming directly from the scattering foil to-
from the scattering foil on the phantom surface. gether with the electrons scattered by the applicator
Electrons that strike the walls of the applicator walls. The dose D(z) in a point at depth I was com-
cause an additional contribution to the dose when puted by integrating the dose contributions to that
scattered back into the beam. Electrons that im- point from the pencil beams confined within the
pinge on the field-defining frame near its edge can treatment field contour. The pencil beams were con-
also scatter back into the field, again adding to the sidered to originate at the phantom’s surface with
dose (Fig. 1). Finally, Bremsstrahlung generated in their axes perpendicular to it. A uniform planar fu-
the accelerator head, the collimating structures and ence at the surface (z = 0) was assumed. For a
in the phantom forms a fourth dose component. treatment field of area A at z = 0 and the depth
axis of interest in (s,_v) = (0.0). DA(z) is given by:

x2 + _I”
A expi .- > ds d_r (1)
202(r)

where 202(;) denotes the mean square radial spread

of the pencil beam dose distribution at depth z. P(z)
represents the central-axis dose values due to elec-
trons in a field with lateral scatter equilibrium [IIS]
on its central axis, normalised to the maximum
dose. These quantities also depend on the energy of
the electron beam at the phantom surface.
For computing the dose due to the electrons scat-
tered by the field-defining frame, the following pro-
cedure was used. The treatment field of area A was
approximated by circle segments extending over an
angle ~40, with their centres on the depth axis of
interest. The dose contribution at depth : on the
axis from a segment with radius R was computed
as (d0/lSO). F(z), with:

F&) = Co(R) - C,(R) - :

for o < ,I < Co(R)/C,(R)
= 0 elsewhere (2)
Fig. I. The different contributions to the dose. (a) Directly in-
cident electrons; (b) applicator wall-scattered electrons: (c)
The functions Co(R) and C,(R) and their param-
frame-scattered electrons. The electron (d) contributes to the
dose in the applicator field but is intercepted when a frame de- eters were determined from central-axis depth-dose
fines a smaller field. curves measured for three circular fields with dif-
398

ferent diameters and for a beam without a frame. axis (infinite field). Q(Z) equals 1 for o(z) = 0 and
These measurements were performed in electron decreases with increasing b(z). From Eqn. (3), val-
beams without using an applicator in order to avoid ues for C(Z) were determined using experimental
distorting effects of wall-scattered electrons. For the Q(z) values. These values are denoted by a”(z) (axis)
field -with area A, the total frame scatter dose on and are used for calculations in all applicators.
the axis at depth z was calculated by summation An a!ternative approach to determine values of
over the segments with d0 = 5 degrees. the Q(Z)parameter employs a set of measured beam
The dose delivered by Bremsstrahlung was de- profiles of a large field [ 13,301.In order to assess the
scribed with a simple exponential function of depth; accuracy of this type of method in depth-dose dis-
the parameters were determined from a measured tribution calculations, we derived a(z) values from
central-axis depth-dose curve [29]. Adding the dose off-axis to on-axis dose ratios at the edge of a 10
due to the directly incident and wall-scattered elec- cm x 10 cm field and used these in a few calcula-
trons, the frame-scatter dose and the photon dose tions. These ts values are denoted by a”(z) (profile).
yielded the total dose in a point on the depth axis. Details of our procedure to obtain aP(z) values are
given elsewhere [6].
Pencil beam parameter values The central-axis percentage depth-dose values for
the largest field in each applicator served as basic
The calculation of depth-dose distributions for ar- data for the calculations. Thus the variation in the
bitrary fields in various applicators requires the fol- wall scatter dose component with applicator aper-
lowing data. Per electron energy one single set of ture was taken into account in P(z). The central-
C(Z)values as a function of depth and for each ap- axis depth-dose curves for the 6 cm x 6 cm, 10 cm
plicator the central axis depth dose curve of the x 10 cm and 14 cm x 14 cm applicator fields were
largest field within the applicator. Values for a(z) measured; for the applicators with intermediate di-
were determined from the central-axis depth-dose mensions, e.g. 8 cm x 8 cm, 6 cm x 14 cm, values
curves measured for a 4 cm x 4 cm and a 10 cm for P(z) were derived by means of linear interpo-
x 10 cm field in the applicator with a 10 cm x 10 lation.
cm aperture after subtraction of the calculated dose
contributions of frame-scattered electrons and pho- Output factor calculation
tons. This applicator was taken because of its in-
termediate dimensions; the 4 cm x 4 cm field was The model of perpendicularly incident Gaussian
considered the smallest square field of clinical in- pencil beams with identical dose distributions as we
terest. The influence of electrons scattered by the employ does not give a realistic description of a
applicator walls on the variation of depth dose beam that contains electrons scattered by the ap-
curves with field size was thus taken into account plicator walls. In order to overcome this limitation
in the a(z) values. without complicating the model, the values for the
At a set of depths the ratios of the two resulting parameters oa(z) and P(z) were derived from mea-
dose values for the small and the large field were sured depth-dose curves. In principle, dose calcu-
calculated; this quantity we denote by Q(Z), thus lations with these c?(z) values do only yield accurate
Q(Z)= ~4W&o(z). A n expression for Q(z) ac- results for the 4 cm x 4 cm field, from which they
cording to the employed model is derived as [9]: were derived. However, we calculated depth-dose
curves for fields with larger and smaller dimensions
Q(Z) = &-{2J02~“(“)exp(- t2/2) dt}2 and it is shown later that the results agree well with
the measurements. On the other hand, a systematic
underestimation was observed in the calculated
assuming the 10 cm x 10 cm field large enough to output factors for fields or parts of fields with a
provide lateral scatter equilibrium on its central dimension smaller than 4 cm. The following simple
 399

correction proved to be adequate for the range of lecting area of this chamber has an elliptical shape
field sizes tested. with axes of 0.3 and 0.8 cm and an electrode sep-
Let the value of Q(z) [Eqn. (3)] at the depth of aration of 0.2 cm [35]. The measured ionisation was
maximum dose in the 4 cm x 4 cm field be Qlim. converted into absorbed dose according to
After the calculation of the dose distribution along Report 35 [18]. Depth-ionisation curves were ob-
the depth axis in an arbitrary field, the value of tained by scanning of the chamber and direct!y
II&)/P(z) [Eqn. (I)] at the depth of maximum dose plotting the signal divided by that of the accelerator
z, was compared with &,,,. When this value was monitor with a pen recorder. The curves were nor-
smaller than Qli,, the output factor was calculated malised to their own maximum value. The accuracy
with DA(z,,,)/P(,,,) = Qlim; in the other case no cor- of such a curve was analysed with respect to posi-
rection was applied. tioning errors of the ionisation chamber relative to
the water surface and to the treatment field contour
aterials and experimental met by repeated measurements. The variation in each
point of the percentage depth-ionisation curve did
Electron beams
not exceed 0.5% of the maximum ionisation value
Dose distributions of electron beams with 6, 8, 10, or 0.05 cm in depth. The maximum ionisation value
12, 14, 17 and 20 MeV nominal energy generated was related to that in a reference field by means of
by a Philips (MEL) SL75-20 linear accelerator were integrated measurements per fixed number of mon-
investigated. Nominal energy refers to the energy itor units. The precision of an output factor deter-
as indicated by the accelerator push buttons; details mination was assessed with regard to variations in
about the scattering and collimation system and the the accelerator output and also to positioning er-
characteristics of the beams can be found elsewhere rors. The maximum difference in the values of the
[8,35]. The applicators have square apertures with measured output factor was 0.5%.
dimensions of 6, 8, 10, 12 and 14 cm and rectan-
gular apertues of 6 x 10, 6 x 14, 8 x 12 and 10
x 14 (cm x cm). Applicators with larger apertures esults and djsc~ssi~~
are also used in our department, but were not in-
cluded in this study because they are of a different Perlcil h>am paramcwrs
construction [S]. Shielding frames in contact with
the phantom surface defined various treatment The values of aa for the beams of 6-20 MeV
fields. These frames consisted of 1 cm of Cerrosafe nominal energy were derived from the measured
(composition 42.5% bismuth, 37.7% lead, 11.3% central-axis dose distributions in the IO cm x 10
tin and 8.5% cadmium) and 0.4 cm of Perspex be- cm applicator with the 10 cm x 10 cm and 4 cm
tween the metal and the phantom surface. The x 4 cm field. Figure 2 shows the polynomials that
frames could be attached to the applicator end; ex- were fitted to the oa(z) data points for four energies
periments were also conducted without an aplica- and that were used in the calculations. The surface
tor. In that case the frames, whose outer dimensions values of a”(z) lie between 0.8 and 1.0 cm, which
were 20 cm x 20 cm, were then positioned on the does not agree with the employed model of parallel
phantom surface with the same distance to the scat- pencil beams that originate at the phantom surface:
tering foil as in the presence of an applicator. i.e. theoretically o(O) = 0. This deviation is partly
due to the influence of electrons scattered by the
Ionisation chamber measurements applicator walls. Because of their oblique incidence,
interception of those electrons by the 4 cm x 4 cm
Measurements were performed in a water phantom frame results in stronger reduction of dose as com-
wit5 a plane-parallel ionisation chamber. The col- pared to a situation without wall scatter (Fig. 1).
400

z 1.6
Y
(II
P E=2OMeV

n
5 1.4
9 80-

Id
5
l.2 Q
E.14 MeV H
1.0 60 -

\
0.6 :_E;6Me;.10.ev
40 -
0.6 -

44 -

0.2 -

I I I I I I 1 I I
OO 2 4 6 8 10
depth km) depth km1

Fig. 2. Pencil beam parameters ea(.z) for the beams with appli- Fig. 3. Central-axis depth-dose curve of the 3 cm x 6 cm field
cator as function of depth and for different electron energies. in the 6 cm x 6 cm applicator. E = 20 MeV; -, measured
along the axis through C; calculated with a”(z) and
This yields smaller Q(Z) values and via Eqn. (3) one frame-scatter correction, m m deviation 1% and 0.2 cm.
then obtains increased aa values (compare with Output factor measured: 1.052; calculated: I .025; , measured
for the 6 cm x 6 cm field. The inset shows the contours of the
Fig. 6).
frame and the point of measurement.
Other effects not included in the model, like
large-angle scattering in the medium, also influence
the empirical ga(z) values. Further, experimental 3 cm x 6 cm field, where the error was 2.7%.
uncertainties should also be considered (see below): Figure 3 shows the strong degradation of the depth-
dose curve for this case (solid line) compared to the
Comparison of dose calculations with measurements curve of the applicator field (dashed line). The cal-
culated depth-dose curve, however, closely approx-
Firstly, a systematic investigation of square and imates the measured one. The combined error in
rectangular fields was carried out. Central-axis per- depth-dose curve and output factor at any depth
centage depth-dose curves and output factors were for all fields remained within 3.5% limits.
measured and calculated for fields of 4 x 4, 6 x As an example for arbitrary field shapes, we show
6, 3 x 6, 3 x 10, 3 x 14 (cm x cm) and those two calculated depth-dose distributions for the
equal to the applicator aperture dimensions. Elec- broad and the narrow part of a field at 14 MeV
tron energies of 6, 10, 14 and 20 MeV were selected electron energy (Fig. 4, closed circles); good agree-
for all these tests; for the 4 cm x 4 cm and appli- ment with the measured curves is obtained. The
cator fields also the beams of 8, 12 and 17 MeV difference between the curves along the axes in
were included in the tests. The calculations were point A and B is striking; in particular the depth of
performed as described above using the aa the 85% point (“therapeutic range”, [18]) in A is 1
parameter values; when a different procedure was cm less than in B. The same calculations for the
used this is mentioned explicitly. Deviations of the electron energies of 6, 10 and 20 MeV showed de-
calculated depth-dose curves from the measured viations in the depth-dose curves of less than 2%
ones nowhere exceeded 0.2 cm in depth or 2% in and 0.2 cm and in the output factors of less than
dose. The maximum difference between calculated 1.5%. In order to test the alternative approach, the
and measured output factors was 2%, except for caiculations were also performed with the aP(z) val-
the extreme case of 20 MeV electron energy and the ues. In the broad part of the field (point B), this
 yieldedexactly the same results as th
with a”(z). In the narrow part at 14
energy, bwever, a total error of 8% in dose at the
depth of the 85% point rsr;u!ted (Fig_ 4, nn9n “k""
circles). For the other energies, similar unaccept-
able deviations were found.
Calculations for another irregular field in a rec-
tangular applicator at 10 eV electron energy also
yielded accurate results (Fig. 5). Note the shift to-
wards the surface of the depth of maximum dose
compared to the applicator field curve; neglection
of this effect would underestimate the output factor
by approximately 5%. For the other energies, the
maximum errors in the calculated depth-dose
depth km)
curves for this field were 2% and 0.2 cm, and in the
Fig. 4. Depth-dose curves for an irregularly shaped field in the output factors 2% (only for 20 MeV: 2.5%).
14 cm x 14 cm applicator. E = 14 MeV; -. measured;
calculated with as(z) and frame-scatter correction, max-
imum deviation 2% and 0.1 cm. Output factor measured in A: Effect oj’applicator-wall scatter
1.028, in B: 1.021. Calculated output factor with a”(z) and
frame-scatter correction in A: 1.025, in B: 1.027; 00 0, cal- To assess the accuracy of our method for electron
culated with r+‘(z) and frame-scatter correction. maximum de- beams with different properties, experiments with-
viation in A: 5% and 0.5 cm; calculated output factor: 1.058.
out applicators at 20 MeV electron energy were
conducted. This energy was chosen because the
variation in depth-dose curves and output factors
with field size was most pronounced at this energy.
The ha(z) parameter values were derived from the
central-axis depth-dose curve of the unshielded
beam and of a 4 cm x 4 cm field, together with the
ratio of the maximum dose values on the central
60 axes of the two fields. The 4 cm x 4 cm field was
defined solely by a frame on the phantom. Again
a 8(z) value larger than zero is observed for z =
0 (Fig. 6).
However, now the uncertainties in the measured
data that enter in the Q(Z) values from which O”(L)
is calculated have to be considered. A positioning
error in depth of the chamber does not change dur-
ing both sets of measurements and only causes a
shift in depth of the Q(Z) values and consequently
of the aa curve. Because of the symmetry of the
depthkm)
frame, a positioning error in lateral direction has a
Fig. 5. Depth-dose curve for an irregularly shaped field in the 8 negligible effect on the outcome. An underestima-
cm x 12 cm applicator, E = 10 MeV: -, measured along
tion of the output factor of the 4 cm x 4 cm field
axis through A; , calculated with #(z) and frame-scatter
due to accelerator instability, however, gives SYs-
correction, maximum deviation 2% and 0.1 cm. Output factor
measured: 0.995: calculated 0.987; -0. measured for the 8 cm tematically too low Q(Z) values, which yields too
x 12 cm field. high aa values at all depths (an overestimation of
402

“E 1.6 - the output factor yields the reverse). Using the up-
0 per and the lower limit of the spread in Q(Z) of
0.5%, two sets of aa were calculated; for Q(Z)
b
> 1 we put 8(z) = 0. Figure 6 shows that the
1.2 -
. deviations from zero of aa for z = 0 to 3 cm are
1.0 - within the experimental uncertainty; in this depth
region, Q(Z) remains close to unity and the aa
values are badly defined. This is in contrast with the
aa curve determined in applicator 10 cm x 10
...’ cm (Fig. 2); there the spread in Q(z) produces a
0.4 - 2’ variation in the values of@(z) of 0.05 cm at most.
??
.*’
.I .*
Due to the applicator-wall scatter effect, overall
0.2 - *.. /
,..* lower values of Q(Z) are obtained which lie in a
I I I I region of slower variation of Q with Q [Eqn. (3)].
OO 2 4 6 8 10
depth (cm)
With the aa values for the uncollimated beam,
the central-axis depth-dose distributions of a 6 x
Fig. U. Pencil beam parameters o”(z) and oP(:) versus depth for
the beam without applicator. E = 20 MeV, -, #(z), de-
6, 3 x 10 and a 3 x 6 (cm x cm) field were cal-
rived from central-axis depth-dose values; the vertical bars in- culated and compared to measurements. The max-
dicate the limits of the u”(7) values due to an uncertainty of 0.5% imum deviations from the curves measured without
in the experimental dai:; OW, C+‘(Z),derived from lateral dose an applicator were 1% and 0.1 cm; the maximum
profiles.
error in the calculated output factors was 1.5%. A
test for an irregular field (Fig. 7) showed excellent
agreement between the measured and calculated
depth-dose curve and a difference of 1% in the out-
put factor. Thus the method also yields accurate
results in complete absence of wall-scattered elec-
trons.
The differences between the oa(z) and aP(z) values
for the beam without applicator at 20 MeV electron
energy are within the given spread in c”(z) (Fig. 6),
which is related to a dose variaton of f 0.5%. Con-
sequently, calculations with @(z) values for the
square and rectangular fields mentioned in the pre-
vious paragraph, including the 4 cm x 4 cm field
and for the field of Fig. 7, yielded similar good re-
sults as with the aa parameter values. The max-
I I I I I I I I ,
imum deviations from the measured depth-dose
OO 2 4 6 8 10 curves were 2% and 0.2 cm and in the output fac-
depth km) tors 1.5%. Calculations with bp(z) at 10 MeV elec-
Fig. 7. Depth-dose curve for an irregularly shaped field in the tron energy yielded results with similar accuracy.
beam without applica = 20 MeV; -, measured along Thus for the uncollimated beams, either method to
the axis through A; calculated with aa(z) and frame- derive a(z) values can be used; the dose calculations
scatter correction. maximum deviation 1% and 0. I cm. Mea-
agree equally well with the measurements. If, how-
sured output factor: 1.051; calculated with aa(z) and frame-scat-
ter correction: 1.042; 000, calculated with aa(z) without ever, the principal aim is to determine the values of
frame-scatter correction, maximum deviation 5% and 0.7 cm; the model parameter, then the aP(z) method is the
calculated output factor: 1.000; m, measured without frame. more stable one, expecially at small depths.
 403

ElfJeer qf, fiance scatter computation method takes such a beam component
into account and could be useful when field shaping
In all preceding calculations, the dose due to elec- T is considered.
trons scattered by the field-defining frame was tak- Only a limited amount of experimental work is
en into account. To show the importance of this required to obtain the necessary data for the pencil
contribution, a depth-dose distribution along the beam model. It comprises of two central-axis depth..
axis in point A of the field of Fig. 7 was also cal- dose distributions in the square applicator with the
culated without the frame scatter dose component medium sized aperture: one for the smallest square
for 20 MeV electron energy (Fig. 7, open circles). field applied in clinical practice and one for the ap-
At the depth of the 85% point, the deviation from plicator field. If the contribution of electrons scat-
the measured curve is 5% or 0.5 cm. This is caused tered by the applicators is considerable, two addi-
by the underestimation of the dose in the first few tional depth-dose measurements are required: one
centimeters; the error in the output factor was 5%. for the smallest and one for the largest square field
Thus for frames in contact with the patient’s skin applicator.
a separate calculation of the dose delivered by the The alternative method that derives values for
frame scatter is required. This is also demonstrated the pencil beam radial width parameter from mea-
by the change in the build-up region of the depth- sured beam profiles should be used with care for
dose curve in Fig. 5. depth-dose curve and output factor prediction. In
beams with a small amount of applicator-wall scat-
ter, good results should be expected. In beams with
Conclusions a non-negligible contribution of electrons coming
from the applicator, however, considerable errors
A method to compute electron beam depth-dose can occur, especially in small fields.
curves and output factors for arbitrary field shapes The effects of electrons scattered by the field de-
has been evaluated, For electron energies between fining frame on depth-dose curves and output fac-
6 and 20 MeV, treatment field dimensions between tors depend on the particular shielding material and
3 and 14 cm and various applicator sizes the on the distance of the frame to the patients’ skin.
method showed sufficient accuracy for clinical prac- Whether or not a frame scatter calculation has to
tice, i.e. the agreement between calculated and mea- be included in the model should be investigated by
sured percentage depth-dose curves was better than the user, as the dose contribution can be of clinical
2% in relative dose and better than 0.2 cm in posi- importance.
tion. The accuracy in computed output factors was
better than 2% compared to measurements, except efeerences
for the extreme case of 20 MeV electron energy and
3 cm field width, where a deviation of 3% occurred. Biggs, P.J.. Boyer. A.L. and Doppke, K.P. Electron dosi-
Treatment beams which contained a considerable metry of irregular fields on the Clinac 18. Int. J. Radiat.
amount of electrons scattered by the applicator Oncol. Biol. Phys. 5: 433-440. 1979.
Bjgrngard. B.E.. Piontek. R.W. and Svensson. G.K. Elcc-
walls as well as beams without such a component
tron scattering and collimation system for a 12 MeV linear
were investigated. The calculations yielded results accelerator. Med. Phys. 3: 152-158, 1976.
with the same accuracy for both cases, which indi- Brahme, A., Lax, 1. and Andreo. P. Electron beam dose
cates the general applicability of the method. Mod- planning using discrete Gaussian beams, mathematical back-
ern systems of electron beam collimation avoid ap- ground. Acta Radiol. Oncol. 20: 147-158, 1981.
Briot, E. and Dutreix, A. Etude experimentale de la colli-
plicator scatter as much as possibe. In intra-oper-
mation des faisceaux d’electrons par un diaphragme dc
ative radiotherapy (IORT), however, closed-wall plomb reglable, J. Radiol. Electrol. 57: 447-454. 1976.
applicator are used that produce an important con- Bruinvis, I.A.D. Physical aspects of electron therapy up to
tribution of scattered electrons. The present dose 20 MeV. Medica Mundi 24: I 19-126. 1979.
404

6 Bruinvis, I.A.D., Van Amstel, A., Elevelt, A.J. and Van der 23 Meyer, J.A., Palta, J.R. and Hogstrom, K.R. Demonstra-
Laarse, R. Calculation of electron beam dose distributions tion of relatively new electron dosimetry measurement tech-
for arbitrarily shaped fields. Phgs. Med. Biol. 28: 667-683. niques on the Mevatron 80. Med. Phys. 11: 670-677, 1984.
1983. 24 Mills, M.D., Hogstrom, K.R. and Almond, P.R. Prediction
7 Bruinvis, I.A.D., Van Amstel, A., Elevelt. A.J. and Van der of electron beam output factors. Med. Phys. 9: 60-68, 1982.
Laarse, R. Dose calculations for arbitrarily shaped electron 25 Mills, M.D., Hogstrom, K.R. and Fields, R.S. Determi-
beams. In: Computed Electron Beam Dose Planning, pp. nation of electron beam output factors for a 20-MeV linear
73-79. Editor: A. Brahme. Acta Radiol. Suppl. 364. 1983. accelerator. Med. Phys. 12: 473-476, 1985.
8 Bruinvis, I.A.D., Heukelom, S. and Mijnheer. B.J. Com- 26 Mini, R. Dosimetry using variable electron tubes. In: Proc.
parison of ionisation medrurements in water and polystyrene World Congress on Medical Physics and Biomedical Engi-
for electron beam dosimetry. Phys. Med. Biol. 30: 1043- neering, p. 27.13. Editors: W. Bleifeld, D. Harder, H.-K.
1053, 1985. Leetz and M. Schaldach. MPBE, Hamburg, 1982.
9 Bruinvis, I.A.D. Electron Beams in Radiation Therapy, pp. 27 Miiller-Runkel. R., Ovadia. J., Borger. F., Culbert, H. and
65-82. Ph.D. thesis, University of Amsterdam, 1987. Rohowsky, B. A shaping device for irregular electron fields
IO Choi, MC., Purdy. J.A., Gerbi, B., Abrath, F.G. and Glas- for the Therac-20 accelerator. Med. Phys. 12: 90-92, 1985.
gow, G.P. Variation in output factor caused by secondary 28 Nair, R.P.. Nair, T.K.M. and Wrede, D.E. Shaped field
blocking for 7-16 MeV electron beams. Med. Phys. 6: 137- electron dosimetry for a Philips SL75/10 linear accelerator.
139. 1979. Med. Phys. 10: 356-360, 1983.
II Clarkson, J. A note on depth doses in fields of irregular 29 Niisslin, F. Ein Rechenverfahren fur die EDV-gestiitzte
shape. Br. J. Radiol. 14: 265-268, 1941. Bestrahlungsplanung in der Therapie mit schnellen Electro-
12 Dutreix, A. and Briot, E. The development of a pencil-beam nen. Habilitationsschrift Medizinische Hochschule, Han-
algorithm for clinical use at the Institut Gustave Roussy. In: nover, 1979.
The Computation of Dose Distributions in Electron Beam 30 Perry, D.J. and Holt, G. A model for calculating the effects
Radiotherapy, pp. 242-270. Editor: A.E. Nahum. Umea of small inhomogeneities on electron beam dose distribu-
University, Umea, 1985. tions. Med. Phys. 7: 207-215, 1980.
13 Edwards, F.H. and Coffey, C.W. A cumulative normal dis- 31 Puel, G. Etude des .faisceaux d’electrons. Application au
tribution model for simulation of electron beam profiles. Int. calcul d’isodoses par ordinateur dans les milieux irradies par
J. Radiat. Oncol. Biol. 5: 127-133, 1979. des electrons de haute energie. Thesis University of Tou-
14 Eyges, L. Multiple scattering with energy loss. Phys. Rev. louse, 198 1.
74: 1534-1535, 1948. 32 Sharma, S.C., Wilson, D.L. and Jose, B. Dosimetry of small
15 Fermi, E. Scattering: The distribution function. In: Cosmic fields for Therac 20 electron beams. Med. Phys. I 1: 697-702,
Ray Theory, pp. 265-268. Editors: B. Rossi and K. Greisen. 1984.
Rev. Mod. Phys. 13: 1940. 33 Sharma, S.C. and Wilson, D.L. Depth dose characteristics
16 Goede, M.R., Gooden, D.S., Ellis, R.G. and Brickner, Jr. of elongated fields for electron beams from a 20-MeV ac-
T.J. A versatile electron collimation system to be used with celerator. Med. Phys. 12: 419-423, 1985.
electron cones supplied with Varian’s Clinac 18. Int. J. Ra- 34 Svensson, H. and Hettinger, G. Influence of collimating sys-
diat. Oncol. Biol. Phys. 2: 791-795, 1977. tems on dose distribution from 10 to 35 MeV electron ra-
17 Hogstrom. K.R., Mills, M.D. and Almond, P.R. Electron diation. Acta Radiol. Oncol. 6: 404409, 1967.
beam dose calculations. Phys. Med. Biol. 26: 445459, 198 1. 35 Van der Laarse, R., Bruinvis, I.A.D. and Nooman, M. Far-
18 ICRU (International Commission on Radiation Units and id wall-scattering effects in electron beam collimation. Acta
Measurements). Radiation dosimetry: electron beams with Radiol. Oncol. 17: 113-124, 1978.
energies between 1 and 50 MeV, Report 35, pp. 46,96-104. 36 Yu, H. The applicability of the method of equivalent
ICRU, Bethesda, Maryland, 1984. squares for photon and electron beams. Phys. Med. Biol. 28:
19 Kahn, F.M., Deibel, F.C., Gerbi, B.J. and Das, I.J. Dosi- 1279-1287, 1983.
metry of irregularly shaped electron fields. Med. Phys., 14: 37 Yudelev, M., Mandelzweig. Y. and Tatcher, M. Relative
473, 1987. output factors of irregularly shaped electron fields. In: Proc.
20 Lax, I. and Brahme, A. Collimation of high energy electron 3rd Annual Meeting ESTRO, Jerusalem, p, 152, 1984.
beams. Acta Radiol. Oncol. 19: 199-207, 1980. 38 Wu, R.K., Wang, W. and El-Mahdi, A.M. Irregular field
21 Loyd, M.D. and Rosen, I. Calculation of output from ir- output factors for electron beams. In: Proc. 9th Int. Conf.
regularly shaped electron beams. Med. Phys. 12: 505, 1985. on the Use of Computers in Radiation Therapy, Scheven-
22 McGinley, P.H., McLaren, J.R. and Barnett, B.R. Small ingen. pp. 453-456. Editors: I.A.D. Bruinvis, P.H. van der
electron beams in radiation therapy. Radiology 13 1: 23 I- Giessen, H.J. van Kleffens and F.W. WittkHmper, 1987.
234, 1979.