You are on page 1of 7

Biomedical Physics & Engineering

Express

PAPER

Empirical formulation for small circular electron fields


To cite this article: Imran Khan et al 2020 Biomed. Phys. Eng. Express 6 065023

View the article online for updates and enhancements.

This content was downloaded from IP address 14.139.45.244 on 12/11/2020 at 13:04


Biomed. Phys. Eng. Express 6 (2020) 065023 https://doi.org/10.1088/2057-1976/abc18b

PAPER

Empirical formulation for small circular electron fields


RECEIVED
8 July 2020
REVISED
Imran Khan1,2 , Sunil Kumar2,∗ , Sushil Kumar3, Tuhin Datta2 and Bipin Singh Koranga4
30 September 2020 1
Department of Physics, Cornell University, Ithaca NY, United States of America
2
ACCEPTED FOR PUBLICATION Department of Physics, Ramjas College, University of Delhi, New Delhi, India
3
15 October 2020 Department of Physics, Hansraj College, University of Delhi, New Delhi, India
4
PUBLISHED
Department of Physics, Kirorimal College, University of Delhi, New Delhi, India

29 October 2020 Author to whom any correspondence should be addressed.
E-mail: skumar@ramjas.du.ac.in

Keywords: electron beam therapy, small circular fields, therapeutic depth, prescription depth, Monte Carlo method, least square fitting,
EGSnrc/BEAMnrc

Abstract
An empirical model for small circular electron fields was developed. This can be of great help in the
treatment planning process for small circular electron fields. A complete dosimetric analysis of the
circular fields defined by electron cutouts diameters (2 cm–9 cm) was done for nominal electron
energies ranging between 6 MeV and 20 MeV using a 3D water phantom and a pin-point ion chamber.
Properties studied included depth dose, in-air inverse-square fall-off, and beam profiles. The Varian
Clianc 2100 C accelerator was modelled, benchmarked and Monte Carlo simulations were performed
using the EGSnrc/BEAMnrc code for the small circular cutouts. A simple exponential model was
found to accurately predict the very important therapeutic depth (90% of Dmax) for the small circular
field size within an accuracy of better than 2 mm in most cases. The model has only two parameters
(d90 and ‘b’). Also, the penumbra widths (90% of the off-axis profiles) of these small circular electron
fields were studied and least square fitted to a simple quadratic model. Full dosimetric profiles of these
small circular electron fields were further studied using the benchmarked Monte Carlo simulations.
This study presents a simple model to predict the very important therapeutic depth (90% of Dmax)
and a recipe to develop such an electron treatment model for any linear accelerator system. Such
predictions can be extremely valuable and time saving prior to treatment planning involving not only
small circular shaped electron fields but also irregularly shaped electron fields.

1. Introduction breast, upper respiratory and digestive tract. Small


electron fields, shaped using circular cutouts made of
The surge in commercial development and applica- Cerrobend are frequently used for treating superficial
tions of the linear accelerators during the seventies for tumors, which are not too deep in the body. Clinically
cancer therapy also led to a surge in use of electrons for it is extremely important to know the depth dose
therapy. Work carried out at some of the premier characteristics and off-axis profiles of these small
institutions during this time, showed that in some circular electron cutouts as they can vary to a great
common situations, there was no alternative treat- extent with the electron energy and the diameter of the
ment to the electron beam therapy. Electrons are circular cutouts [3]. A lot of laborious measurements
charged particle with a finite range and therefore are in this regard may need to be performed before
suitable for treatment of superficial and shallow deciding upon the right electron energy and the
targets. Electron beam therapy offers some distinct circular cutout size for a particular case. Previous
advantages over megavoltage photon beam therapy studies have been performed in this respect to get
[1, 2]. Electrons deliver a reasonably uniform dose Output Factors using the Monte Carlo method [4].
from the surface to a specific depth, after which the Prediction of Output Factors have also been obtained
dose falls off rapidly to a near zero value unlike the using a parametric form [5] and pencil beam algorithm
megavoltage photon beams. Electrons are useful in [6]. Several studies have been done on measuring
treating cancer of the skin and lips, head and neck, dosimetric properties of small electron fields.

© 2020 IOP Publishing Ltd


Biomed. Phys. Eng. Express 6 (2020) 065023 I Khan et al

Figure 1. Circular Cerrobend Electron Cutouts of diameter ranging from 2 cm to 9 cm.

However most of the studies are missing on estimating 2. Methods


a very important parameter for electron treatment that
is the therapeutic depth (90% of Dmax). The ther- A set of circular electron cutouts, made of Cerrobend
apeutic depth, which is the 80% to 90% of the depth of with physical diameters from 2 cm to 9 cm were
the depth dose curve coming after Dmax is the most constructed in the usual clinical manner (figure 1). A
useful range for any electron field. The therapeutic full dosimetric analysis of the circular fields defined by
depth (90% of Dmax) is given by the depth of the 90% each cutout was done for nominal electron energies
depth dose coming after Dmax, for that particular field ranging from 6 MeV to 20 MeV for a Varian Clinac
size. This is very important as the dose decreases 2100 C. The dosimetric data was collected using a
abruptly beyond the 90% dose level after Dmax. The pinpoint chamber and 3D water phantom as the same
therapeutic depth (90% of Dmax) and the required equipment/process was used in the commissioning
electron energy must be chosen very carefully. In this process of the linear accelerator. Properties studied
study we have characterized the dosimetric properties included depth dose, in-air inverse-square fall-off, and
of small electron cutouts with diameter ranging from beam profiles. A linear accelerator model was built in
2 cm to 9 cm for a Varian Clinac 2100 C. We have EGSnrc/BEAMnrc [7] for the Varian Clinac 2100 C.
arrived at a simple analytical model, which using a The geometry of the linear accelerator (position of
single curve fitted coefficient : ‘b’ can be used for scattering foil, x-y jaws, applicators etc) was obtained
predicting the very important: therapeutic depth (90% directly from Varian Medical Systems of Palo Alto,
of Dmax) for small circular cutouts made of Cerro- California. After appropriate benchmarking, these
bend. Also, we have measured and fitted equations to properties were further studied with Monte Carlo
the 90% field width at SSD of 110 cm, which can be model of the accelerator developed using the EGSnrc/
very useful to decide the energy and cutout size based BEAMnrc code [7]. From the physical data and
on the intended target coverage area. The empirical simulations, the following parameters were deter-
formation developed in this study can be used mined for each cutout/energy combination: R50, R90,
effectively for deciding the right electron energy and widths of 90% and 95% isodose surfaces, effective
the right circular cutout diameter after deciding upon SSD, and dose output factor. Selected parameters were
the right therapeutic depth (90% of Dmax). Similar then correlated with collimator size to arrive at an
measurements may be performed for any linear exponential model, which accurately predicts the
accelerator system to use the simple exponential therapeutic depth (90% of Dmax). For projected
expression with machine specific parameters includ- circular cutout sizes smaller than 2 cm, the measure-
ing the fitting parameter ‘b’. This approach can be used ments are very hard to perform, therefore the results
for making optimal and swift treatment planning for from benchmarked Monte Carlo simulations were
cases involving small circular electron fields and used to fill in the points in order to develop the
shallow targets. complete empirical model. This set of crucial data

2
Biomed. Phys. Eng. Express 6 (2020) 065023 I Khan et al

Figure 2. 90% Depth (y) beyond Dmax versus projected circular cutout diameter (x) for SSD of 100 cm. For x<2 cm, the data points
were generated through benchmarked Monte Carlo accelerator model.

Table 1. Values of fitted parameters evaluated using least square regression analysis.

Parameters Energy=6 MeV Energy=9 MeV Energy=12 MeV Energy=16 MeV Energy=20 MeV

d90 1.846 3.005 4.247 5.475 6.061


b 0.7517 0.5086 0.4105 0.3536 0.3759

points (x<2 cm) are necessary to arrive at the Table 1 shows the values of the least square fitted
functional form of equation (1). parameter ‘b’, d90 for all the nominal electron energies
for the therapeutic depth (90% of Dmax). These values
3. Results can be easily obtained for any particular linear accel-
erator by performing measurements for a few circular
The following simple exponential model (equation (1)) electron cutouts and a few Monte Carlo simulations.
had been found to accurately predict the commonly used The parameter ‘b’ depends on the beam energy and
the therapeutic depth (90% of Dmax) for the circular has been least square fitted to a quadratic model (cubic
field size with projected diameter x (at SSD = 100 cm), term coefficient is close to zero) as shown in figure 3
within an accuracy of better than 2 mm in most cases. with coefficient of determination : R2=0.998. This
model can be developed for any accelerator system and
y = d 90[1 - e-b.x ] (1)
used to predict the ‘b’ parameter for a particular
Here y is the 90% depth for the circular cutout with energy to be used the main modeled equation (1) for
projected diameter: x, d90 is the 90% therapeutic or calculating the therapeutic depth (90% of Dmax).
prescription depth (90% of Dmax) for the open cutout Penumbra widths, which are defined as the 90%
(10×10 cm2 field), which is normally measured field width at the depth of maximum dose were eval-
during the commissioning process. This d90 can also uated and are presented in figure 4. These are plotted as
be estimated approximately by E/3.2 cm, where E is a function of electron cutout size (x) and nominal elec-
the most probable energy in MeV of the electron beam tron energy at 110 cm SSD. The data was least square
at the patient/phantom surface [8]. Figure 2 shows
curve fitted to set of quadratic expressions (shown in
the plot of measured values and also those obtained
figure 4) for each electron energy. The coefficient of
from Monte Carlo modelling for projected diameter:
determination (R2 value), which is a measure of how
x<2 cm. The predicted curves are plotted as well by
curve fitting the functional form using least square good the fitting [10] has been calculated and in almost all
method for electron beam energies ranging from cases are very close to the ideal number of one.
6 MeV to 20 MeV. The only real unknown in this Figure 5 shows the measured and Monte Carlo
model here is the parameter ‘b’, to be determined generated data for 12 MeV central axis, percent depth
using least curve fitting. The coefficient of determina- doses for all the small circular cutouts of diameters
tion for the curve fitting (R2 value) was close to 0.99 for ranging from 2 cm to 10 cm. Figure 6 shows a compar-
all the electron energies [9]. ison of the measured and the Monte Carlo generated

3
Biomed. Phys. Eng. Express 6 (2020) 065023 I Khan et al

Figure 3. The parameter ‘b’ of the equation (1) as a function of electron energy along with a least square fitted quadratic model for
predicting it for any console energy: E.

Figure 4. Measured and fitted quadratic equations for 90% penumbra widths at 110 cm SSD.

data for off-axis profiles of 9 MeV electron beams circular sizes of 2 cm and 3 cm is not as great as the
again for all the circular cutouts of diameter ranging agreement for field sizes higher than 3 cm. This slight
from 2 cm to 10 cm. The simulations were done using disagreement may be due the uncertainties involved in
100 million histories to give error bars of sizes less than measurement for very small field sizes. The bench
±1%. The simulation time taken for each cutout was marked Monte Carlo data for percent depth dose may
around 2 to 3 hours on an Intel i5-4570@3.2 GHz pro- be trusted more in these cases.
cessor with 4GB RAM.
The Monte Carlo method needs to be able to predict 4. Discussion and conclusions
a dose distribution in agreement with the measurement
within clinically acceptable tolerance level [11, 12]. A The formulation developed in this work accurately
commonly accepted accuracy level is ±2% or ±2 mm as describes the clinical dosimetric properties of small
stated in ICRU Report 42. [13]. Great agreements (dif- circular electron fields. It can be used to select the
ferences less than 2%) between the measurements and circular cutout diameter needed to provide the desired
the Monte Carlo data have been found in both the cases coverage (using 90% field widths) at the very impor-
studied in figures 5 and 6. The agreement for very small tant therapeutic depth (90% of Dmax), beyond which

4
Biomed. Phys. Eng. Express 6 (2020) 065023 I Khan et al

Figure 5. Measured and Monte Carlo data for 12 MeV, central axis percent depth dose profiles.

Figure 6. Measured and Monte Carlo data for 9 MeV off-axis profiles for circular cutouts with diameters ranging from 2 cm to 5 cm.

the dose falls off very rapidly. This can be done without and energies have been narrowed down, the fitted
relying on usually used approximate rules i.e. estimat- equations of the measured data for penumbra widths
ing it as E/3.2 cm, where E is the most probable energy can be used to close the search for the right cutout and
in MeV of the electron beam at the patient/phantom the right energy. Modelled equation (1) has only two
surface [8] or trying various combinations of energy parameters that of d90 for the open 10×10 field
and cutout size in a treatment planning system. This (which is usually measured during the commissioning
formulation can greatly help in providing adequate process) and the independent parameter ‘b’. The pos-
dose at the right target depth under the skin and have sible values of parameter ‘b’ (shown in table 1) were
intended target coverage. determined using the empirical data and nonlinear
Equation (1) can be effectively used for predicting regression method [9]. It may be noted that the para-
the depth (90% of Dmax), represented as: y for small meter ‘b’ does not vary much for higher energies ran-
circular cutouts with projected diameter: x and greatly ging from 12 MeV to 20 MeV. Parameter ‘b’ may be
help in the decision of choosing the right circular cut- related to the scattering power of the electron beam,
out size and energy. Once a set of relevant cutout sizes which is higher for lower energies. This parameter can

5
Biomed. Phys. Eng. Express 6 (2020) 065023 I Khan et al

easily be obtained by doing measurements for all the ORCID iDs


available circular cutouts and filling in the points for
cutout diameter for x<2 cm, using the benchmarked Imran Khan https://orcid.org/0000-0002-4058-8561
Monte Carlo simulations. Finally, this ‘b’ parameter Sunil Kumar https://orcid.org/0000-0001-8240-0945
can be predicted using the least curve fitted quadratic
function shown in figure 3. A similar functional form References
can be easily developed for any accelerator system.
[1] Khan F M, Doppke K P, Hogstrom K R, Kutcher G J, Nath R,
The Monte Carlo method relies on fundamental Prasad S C, Purdy J A, Rozenfeld M and Werner B L 1991 The
physics principles governing interaction of particles in international journal of medical physics and practice 18 73
media and has been proved to be capable of accurately AAPM Report No. 32 (Clinical Electron Dosimetry)
[2] Hogstrom K R and Almond P R 2006 Review of electron beam
modeling electron beams in radiation therapy [11, 12].
therapy physics Phys. Med. Biol. 51 R455–89
A lot of improvement has been attained in this regard [3] Khaledy N, Arbabi A and Sardari D 2011 The effects of cutouts
[9]. Final choice of the circular cutout size and energy on output, mean energy and percentage depth dose of 12 and
14 MeV electrons J. Med Phys. 36 213–9
can be used as input into the benchmarked Monte
[4] Turian J V, Smith B D, Bernard D A, Griem K L and Chu J C
Carlo model to get the full dosimetric properties of the 2004 Monte Carlo calculations of output factors for clinically
cutout: percent depth dose (figure 5) and off axis pro- shaped electron fields J Appl Clin Med Phys. 2004 Spring 5
files (figure 6) at the chosen energy. Processing time 42–63
[5] Akino Y, Zhu T C and Das I J 2015 Parameterization of
may be saved in performing benchmarked Monte electron beam output factor Phys Med. 31 420–4 Epub
Carlo simulations by using phase-space files. A phase- [6] Mills M D, Hogstrom K R and Almond P R 1982 Prediction of
space file is a collection of data specifying properties electron beam output factors Med. Phys. 9 60–8
[7] Kawrakow I, Mainegra-Hing E, Rogers D W O, Tessier F and
such as the position, direction, energy and the type for Walters. B R B 2017 The EGSnrc Code System: Monte Carlo
each particle crossing a plane defined as a phase-space simulation of electron and photon transportTechnical Report
plane. These phase-space files are usually compatible PIRS-701National Research Council Canada (https://nrc-
cnrc.github.io/EGSnrc/doc/pirs701-egsnrc.pdf)
with the International Atomic Energy Agency (IAEA)
[8] Nordic Association of Clinical Physics 1980 Procedures in
format for clinical electron energies just above the external radiation therapy dosimetry with electron and photon
jaws. These are provided for some of the new systems beams with maximum energies between 1 and 50 MeV Acta
such as the Trubeam Linacs from Varian. Complete Radiol. 11 55
[9] Draper N R and Smith H 1998 Applied Regression Analysis
studies of small electron fields using such a phase space (Hoboken, NJ: Wiley-Interscience) (https://wiley.com/en-
file approach has been demonstrated [14]. This recipe in/Applied+Regression+Analysis,+3rd+Edition-p-
may be followed for any of the commercial linear 9781118625682) 978-0-471-17082-2
[10] Sachse K N and du Plessis F C P 2019 Improved Monte Carlo
accelerator, thereby greatly cutting down the treat- clinical electron beam modelling’. Phys Med. 66 36–44
ment planning time and effort without compromising [11] Ma C M and Jiang S B 1999 Monte Carlo modelling of electron
on the optimization of the planning process. Similar beams from medical accelerators Phys. Med. Biol. 44 R157–89
[12] Chetty I J et al 2007 Report of the AAPM Task Group No. 105:
studies and modelling may be designed to be carried Issues associated with clinical implementation of Monte
out for other symmetrically shaped small electron cut- Carlo-based photon and electron external beam treatment
outs such as triangular, square or rectangular shaped. planning Med. Phys. 34 4818–53
[13] Thwaites D 1987 Use of computers in external beam
This study was specially carried for small circular elec-
radiotherapy procedures with high-energy photons and
tron cutouts but can be useful for irregularly shaped electrons ICRU Report 42 (https://doi.org/10.1016/S0936-
electron fields as well. The benchmarked Monte Carlo 6555(05)80223-7)
models (tested for small circular cutouts) may be [14] Rodrigues A, Sawkey D, Yin F-F and Qiuwen W 2015 A Monte
Carlo simulation framework for electron beam dose
relied upon to get the full dosimetric properties of any calculations using Varian phase space files for TrueBeam
irregularly shaped field. Linacs Med. Phys. 42 2389

You might also like