ARTICLE IN PRESS

Applied Radiation and Isotopes 67 (2009) 266–271

Contents lists available at ScienceDirect

Applied Radiation and Isotopes

journal homepage: www.elsevier.com/locate/apradiso

Monte Carlo simulation of X-ray spectra and evaluation of filter effect using
MCNP4C and FLUKA code
R. Taleei a, M. Shahriari b,
a
Department of Radiation Medicine, Shahid Beheshti University, Tehran, Iran
b
Department of Radiation Application, Shahid Beheshti University, Tehran, Iran

a r t i c l e in f o a b s t r a c t

Article history: The general-purpose MCNP4C and FLUKA codes were used for simulating X-ray spectra. The electrons
Received 19 July 2008 were transported until they slow down and stop in the target. Both bremsstrahlung and characteristic
Received in revised form X-ray production were considered in this work. Tungsten/aluminum combination was used as target/
6 October 2008
filter in the simulation. The results of two codes were generated in 80, 100, 120 and 140 kV and
Accepted 16 October 2008
compared with each other. In order to survey filter effect on X-ray spectra, the attenuation coefficient of
filter was calculated in 120 kV. More details of filter effect have been investigated. The results of
Keywords: MCNP4C and FLUKA are comparable in the range of bremsstrahlung spectra, but there are some
X-ray spectra differences between them especially in specific X-ray peaks. Since the specific peaks have not significant
Monte Carlo simulation
role on image quality, both FLUKA and MCNP4C codes are fairly appropriate for production of X-ray
Filter effect
spectra and evaluating image quality, absorbed dose and improvement in filter design.
X-ray tube design
MCNP4C & 2008 Published by Elsevier Ltd.
FLUKA

1. Introduction
X-ray machines are one of the most important facilities which
are utilized in radiography and radiotherapy. Knowledge of the
X-ray energy spectrum of a diagnostic X-ray tube is essential for
evaluating image quality and absorbed dose to the patient
(Sanborg et al., 1994). X-ray spectra have been determined by
measuring the X-ray output experimentally. Experimental mea-
surement of X-ray spectra requires special equipment which is
available only in a limited number of laboratories (Fewell and
Shuping, 1978; Fewell et al., 1981; Antonuk et al., 1997; Wilkinson
et al., 2001). Fewell et al. measured X-ray spectra with different
target/filter combinations for over two decades and have pub-
lished several measured spectra (Fewell and Shuping, 1978; Fewell
et al., 1981). However, the measurement of X-ray spectra is
impractical for a clinical X-ray machine and specialized equip-
ment is required which is available only in some laboratories.
Different methods have been developed for numerically predic-
tion of X-ray spectra. These can be divided into three categories:
empirical models (Fewell and Shuping, 1978; Fewell et al., 1981;
Boone et al., 1997; Boone and Seibert, 1997), semi-empirical
models (Boone, 1988; Birch and Marshall, 1979; Tucker et al.,
1991; Blough et al., 1998) and Monte Carlo modeling (Kulkarni
 Corresponding author. Fax: +98 21 22431780.
E-mail address: m-shahriari@sbu.ac.ir (M. Shahriari).
and Supe, 1984; Acosta et al., 1998; Bhat et al., 1999; Verhaegen
et al., 1999; Ng et al., 2000; Ben Omrane et al., 2003; Verhaegen
and Castellano, 2002). Although, purely empirical and semi-
empirical models remain the fastest methods for X-ray spectra
prediction, the models proposed so far still have limitations which
prevent their adoption for a large range of applications (Ay et al.,
2004a). The use of the Monte Carlo method to simulate radiation
transport has become the most accurate means of predicting the
X-ray spectra even in complex geometries owing to more accurate
physics modeling. Moreover, the method tracks the evolution of
all secondary particles (and their descendants) generated by
primary electrons. For the purpose of Monte Carlo simulation of X-
ray spectrum, some authors have used self-written or in house
developed computer codes (Kulkarni and Supe, 1984; O’Meara
et al., 1998), while others have used public domain general-
purpose Monte Carlo codes such as EGS4 (Bhat et al., 1998, 1999;
Ben Omrane et al., 2003), MCNP (Verhaegen et al., 1999; Mercier
et al., 2000) and ITS (Ng et al., 2000). Nevertheless, the prediction
of X-ray spectra using the Monte Carlo method is time consuming
compared to empirical and semi-empirical models. Few works are
available on X-ray spectra production, using general-purpose
Monte Carlo codes especially new codes like FLUKA. On the other
hand, these codes have many advance features that can be used
for more complicated problems, like simulating CT images of
complex phantom geometries and dosimetric calculations. Indeed
these codes have many variance reduction capabilities, which can
be used for time consuming simulations such as dose calculation

0969-8043/$ - see front matter & 2008 Published by Elsevier Ltd.

doi:10.1016/j.apradiso.2008.10.007
 ARTICLE IN PRESS

R. Taleei, M. Shahriari / Applied Radiation and Isotopes 67 (2009) 266–271 267

in fluoroscopy. The first step for testifying the multipurpose codes

in such problems is to estimate their accuracy on X-ray spectra
production.
In our previous work, MCNP4C Monte Carlo code has been
used to simulate the diagnostic radiology and mammography
X-ray tube (Ay et al., 2004b). In this work, we used FLUKA Monte
Carlo code to produce X-ray spectra and we compared it with
MCNP4C and experimental results. The effect of Al filter on X-ray
spectra has been investigated in more detail.

2. Material and methods

2.1. The MCNP4C code

MCNP4C is a general-purpose three-dimensional Monte Carlo

code that can be used for neutron, photon and electron or coupled
neutron/photon/electron transport (Briesmeister, 2000). For
photon transport, the code takes into account incoherent and
coherent scattering, the possibility of fluorescent emission after
photoelectric absorption and bremsstrahlung. The continuous
slowing down approximation energy loss model is used for
electron transport. To follow an electron through a significant
energy loss, the MCNP4C code breaks the electron’s path into
many steps. These steps are chosen to be long enough (with valid
multiple scattering theories) to encompass many collisions but Fig. 1. Geometry of X-ray tube.
short enough so that the mean energy loss in any one step is small
(to ensure that approximations necessary for multiple scattering
theories are satisfied). Except for the energy loss and straggling 2.3. The specification of X-ray tube
calculation, the detailed simulation of the electron history takes
place in the sampling of the substeps. The Goudsmit–Saunderson The geometry of X-ray tube has been implemented accurately
theory is used to sample from the distribution of angular in the simulation. The angle of electron’s departure from the
deflections, in a way that the direction of the electron can change cathode is precisely considered in order to get 1.2 mm focal spot.
at the end of each substep. For electron transport, MCNP4C The anode is made from tungsten with an angle of 121. The
addresses the sampling of bremsstrahlung photons at each aluminum filter thickness is 1.2 mm (Fig. 1). When the electron
electron substep. The table of production probabilities is used to beam hits the tungsten target, X-ray photons are generated. The
determine whether a bremsstrahlung photon will be created energy of electron beam was chosen to be 80, 100, 120 and
(Hughes, 1997). The el03 database also includes characteristic k X- 140 keV. The electrons and photons have been transported inside
ray production probabilities. The el03 evaluation is derived from the target and filter and the bremsstrahlung and specific X-ray
the ITS 3.0 code system (Halbleib et al., 1992). spectrum have been produced. The spectra of these photons have
been calculated by track length method in both Monte Carlo
2.2. The FLUKA code codes.

FLUKA uses an original multiple Coulomb scattering transport

algorithm for electrons, which gives the correct lateral displace-
ment even near the boundaries (Ferrari et al., 1991). In each step
of the multiple scattering algorithms, the discrete event cross
sections of the continuous energy loss are considered. The
bremsstrahlung differential cross sections of Seltzer and Berger
(Ferrari et al., 1991; Seltzer and Berger, 1985, 1986) have been
extended, and the angular distribution of bremsstrahlung photons
is sampled accurately. The Landau–Pomeranchuk–Migdal sup-
pression effect (Landau and Pomeranchuk, 1953; Migdal, 1956,
1957; Landau and Pomeranchuk, 1965) and the Ter-Mikaelyan
(1954) polarization effect in the soft part of the bremsstrahlung
spectrum are also implemented. The present lowest transport
limit for electrons is 1 keV. Although in high-Z materials the
Moli‘ere multiple scattering model becomes unreliable below
20–30 keV, a single-scattering option is available which allows to
obtain satisfactory results in any material also in this low-energy
range. The characteristic X-rays in FLUKA are optionally activated
either by the default physics selected for the simulation or by the
card EMFFLUO. The FLUKA code will take into account all detailed
interactions of six K and L single subshells. The fluodt.dat includes
the cross sections of characteristic X-rays and other data, required
for low-energy electron–photon transport (Fasso et al., 2003).
2.4. The effect of Al filter
In order to evaluate the effect of Al filter, the X-ray spectrum is
calculated before and after the Al filter by both codes. For this
purpose the X-ray spectrum of 120 kV has been treated as a point
source in the center of spherical Al filter. With this method the
calculation time has been reduced to several minutes. In order to
scrutinize the Al filter effect in more detail, photon sources with
energies from 10 to 140 keV with 10 keV steps has been
considered. The values of collided and uncollided photons as well
as total penetrating photons have been calculated with both
codes.
3. Results and discussions
One of the most important parameters influencing the quality
of X-ray spectrum is filtration. The produced X-ray profiles before
and after the filter using FLUKA code are shown in Figs. 2–3. The
corresponding results of MCNP simulation are similar to the
FLUKA results. The X-ray spectra show continuous and character-
istic X-rays spectra of tungsten. The effect of Al filter in
 ARTICLE IN PRESS

268 R. Taleei, M. Shahriari / Applied Radiation and Isotopes 67 (2009) 266–271

5.00E-06

4.50E-06

4.00E-06

3.50E-06
#/cm2/source particle

3.00E-06 80 kV
100 kV
120 kV
2.50E-06 140 kV
2.00E-06

1.50E-06

1.00E-06

5.00E-07

0.00E+00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Energy (kV)

Fig. 2. X-ray profiles before Al filter simulated by FLUKA in 80, 100, 120 and 140 kV.

1.20E-06

80 kV
100 kV
1.00E-06 120 kV
140 kV
#/cm2/source particle

8.00E-07

6.00E-07

4.00E-07

2.00E-07

0.00E+00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Energy (kV)

Fig. 3. X-ray profiles after Al filter simulated by FLUKA in 80, 100, 120 and 140 kV.

attenuation of low-energy X-rays is also obvious by comparing
Figs. 2 and 3.
In order to compare the results of simulation for both codes,
and the X-ray profiles before and after filter, 120 kV profiles have
been chosen. In Fig. 4 the X-ray profiles of both codes before Al
filter with 0.5 keV energy bins are sketched. An extreme difference
in characteristic X-rays results is seen below 15 keV, which is
mainly due to difference between cross section libraries of two
codes. FLUKA simulation gives two peaks near 10 keV, while
MCNP4C gives only one peak. Of course the height of MCNP4C
peak is greater than FLUKA peaks. Fig. 5 compares X-ray profiles of
both codes after Al filter in 120 kV with experimental data.
According to experimental data resolution, 2 keV energy bins has
been chosen. For this purpose four characteristic X-ray peaks have
been reduced to two peaks. The specific X-rays have different
height due to the difference between the algorithms and cross
section libraries of the codes. According to the results, FLUKA
underestimates characteristic X-rays comparing to the experi-
mental data with 11% for both peaks, while MCNP overestimates
them with 25% and 11% for the first and second peak, respectively.
Table 1 gives the ratio of characteristic X-rays calculated by MNCP
and FLUKA relative to the experimental data in four energies.
The effectiveness of Al filter to attenuate low-energy photons is
very more than high-energy photons. Therefore the patient dose is
reduced. In Fig. 5 the results of the bremsstrahlung photon
spectrum of two codes show a great agreement with the
experimental results especially at higher energies after the
characteristic peaks.
For evaluating Al filter effect on X-ray spectrum, the ratio of
profile after filter to its value before filter for both codes in 120 kV
has been calculated by both codes. The ratio indicates the
efficiency of Al filter quantitatively. The results have been shown

#/cm2/source particle
0.08
Experimental Data
0.07 MCNP
Averaged Number of Photons
FLUKA
0.06
0.05
0.04
0.03
0.02
0.01
0 Carlo simulation is the most accurate method for predicting X-ray
0 20
Fig. 5. Comparison of X-ray profiles of FLUKA and MCNP4C simulations after Al
filter and experimental data in 120 kV.
in Fig. 6, and show a great agreement between two codes. The
attenuation below 12 keV is very high and approximately all
photons are absorbed in the filter. The penetration is increased
steeply by photon energy, and reaches to about 90% at 50 keV. The
penetration growth continues further with a smooth slope until it
reaches about 95% at 120 keV. In order to scrutinize the filter
effect in more detail photon energies from 10 to 140 keV in 10 keV
steps with a 1.2 mm Al filter has been simulated. Table 2 shows
the results for both codes. The emitted photon either penetrates or
is absorbed in the filter. The penetrating photon either collides or
passes the filter without collision. The results of two codes show
great agreement with more than 0.5% accuracy. Since the photo-
electric is the dominant process in lower energies, the most of
collided photons are absorbed in this range. The cross section of
photoelectric effect decreases with increase in photon energy.
Near 60 keV the photoelectric cross section is equal to Compton
scattering cross section for Al. Therefore as seen in Table 2 the
penetrating collided photon are approximately equal to the
absorbed collided photon near 60 keV. For example the MCNP
results are 3.89% and 3.4%, respectively. In Fig. 6 the results are a
little bit different from 35 to 85 keV for both codes. According to
Table 2 the disagreement is not caused by photon attenuation in
ARTICLE IN PRESS
R. Taleei, M. Shahriari / Applied Radiation and Isotopes 67 (2009) 266–271 269
9.00E-06 MCNP Before Al Filter
Fluka Before Al Filter
8.00E-06
7.00E-06
6.00E-06
5.00E-06
4.00E-06
3.00E-06
2.00E-06
1.00E-06
0.00E+00
0 10 20 30 40 50 60 70 80 90 100 110 120
Energy (KeV)
Fig. 4. X-ray profiles before Al filter simulated by FLUKA and MCNP4C in 120 kV.
Al filter. The main reason is the difference between the codes in
producing X-rays especially characteristic X-rays, which penetrate
without collision. The other probable reason could be the effect of
scattered photons from higher energies.
4. Conclusion
Monte Carlo simulation of X-ray spectra has been used
extensively in different medical imaging applications including
assessment of image quality, optimization of system design and
absorbed dose calculation. Since the experimentally measure-
ments of X-ray spectra are not practical in many cases, Monte
40 60 80 100 120 spectra. This study has employed the FLUKA and MCNP4C
Energy (keV) general-purpose Monte Carlo codes for generating X-ray spectra.
The effect of Al filer on X-ray spectra and attenuation of X-rays
has been determined. At lower energies the photons are almost
absorbed by the filter, which will reduce the patient dose. At
higher energies the penetration increases to more than 99%. These
calculations can be properly used in filter design, and includes
more detail and accurate estimation with respect to half value
layer (HVL) calculations. Although uncollided penetrating photons
are responsible for filter attenuations, the HVL gives an estimation
of total penetrating photons. From Table 2 in 120 keV 99.6% of
photons penetrate the filter, 95.6% of which are uncollided. This is
in agreement with the results of Fig. 6, which indicates 95.5% of
120 keV photons are penetrating the filter. The results of MCNP4C
and FLUKA are comparable in the range of bremsstrahlung
spectra, but there are some differences spatially in specific X-ray
peaks. The height of MCNP4C specific X-ray peaks are about 1.5
times of their respective values in FLUKA results. The FLUKA
specific X-ray peaks underestimate 11% with respect to experi-
mental data. In the range of radiography X-ray, specific peaks do
not play an important role in image quality or dose calculations.
Therefore both FLUKA and MCNP4C codes are fairly appropriate in
simulating X-ray tubes’ profiles for further investigation of image
quality and enhanced filter design. They can also be used for
three-dimensional geometries and provide simulations with
different phantoms, or shielding problems. None of the codes
show great capability for X-ray simulations in respect to the other
code. The difference between the codes is minor and related to the
 ARTICLE IN PRESS

270 R. Taleei, M. Shahriari / Applied Radiation and Isotopes 67 (2009) 266–271

Table 1
Ratio of characteristic X-rays calculated by MNCP and FLUKA relative to the experimental data in four energies.

E (kV) First peak Second peak

MCNP/experiment FLUKA/experiment MCNP/experiment FLUKA/experiment

80 1.13E+00 8.38E01 1.02E+00 7.92E01

100 1.25E+00 8.94E01 1.12E+00 8.98E01
120 1.25E+00 8.90E01 1.11E+00 8.89E01
140 1.25E+00 9.05E01 1.12E+00 9.06E01

100.00
MCNP
90.00 FLUKA

80.00
Relative Filtar Attenuation (%)

70.00

60.00

50.00

40.00

30.00

20.00

10.00

0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Energy (keV)

Fig. 6. Ratio of the X-ray profile after filter to its value before filter simulated by FLUKA and MCNP4C in 120 kV.

Table 2 References
X-ray attenuation by Al filter simulated by FLUKA and MCNP4C.

Acosta, E., Liovet, X., Coleoni, E., Riveros, J.A., Salvat, F., 1998. Monte Carlo
E (keV) Collided photons (%) Uncollided photons (%) Total (%)
simulation of X-ray emission by kilovolt electron bombardment. J. Appl. Phys.
83, 6038–6049.
FLUKA MCNP FLUKA MCNP FLUKA MCNP
Antonuk, L.E., et al., 1997. Empirical investigation of the signal performance of a
high-resolution, indirect detection, active matrix flat-panel imager (AMFPI) for
10 2.47E05 2.00E05 0.0216 0.0199 0.0216 0.0200 fluoroscopic and radiographic operation. Med. Phys. 24, 51–70.
20 0.37 0.38 34.76 35.51 35.13 35.89 Ay, M., Sarkar, S., Shahriari, M., Sardari, D., Zaidi, H., 2004a. Comparative
30 1.80 1.83 72.10 72.60 73.90 74.43 assessment of different computational models for generation of X-ray spectra
40 2.96 2.96 85.20 85.59 88.16 88.55 in diagnostic radiology and mammography. In: Proceedings of the IEEE
50 3.59 3.59 90.30 90.50 93.89 94.09 Nuclear Science Symposium and Medical Imaging Conference, October 19–22,
60 3.91 3.89 92.60 92.70 96.51 96.60 Rome.
70 4.05 4.03 93.80 93.85 97.85 97.88 Ay, M., Shahriari, M., Sarkar, S., Adib, M., Zaidi, H., 2004b. Monte Carlo simulation of
80 4.11 4.10 94.48 94.49 98.59 98.60 X-ray spectra in diagnostic radiology and mammography using MCNP4C. Phys.
90 4.12 4.09 94.86 94.93 98.98 99.02 Med. Biol. 49, 4897–4917.
100 4.10 4.08 95.19 95.21 99.29 99.29 Ben Omrane, L., Verhaegen, F., Chahed, N., Mtimet, S., 2003. An investigation of
110 4.06 4.02 95.41 95.44 99.47 99.47 entrance surface dose calculations for diagnostic radiology using Monte Carlo
120 3.99 3.88 95.60 95.69 99.59 99.59 simulations and radiotherapy dosimetry formalisms. Phys. Med. Biol. 48,
130 3.96 3.92 95.73 95.75 99.69 99.68 1809–1824.
140 3.91 3.99 95.84 95.75 99.75 99.75 Bhat, M., Pattison, J., Bibbo, G., Caon, M., 1998. Diagnostic X-ray spectra: a
comparison of spectra generated by different computational methods with a
measured spectrum. Med. Phys. 25, 114–120.
Bhat, M., Pattison, J., Bibbo, G., Caon, M., 1999. Off-axis X-ray spectra: a comparison
of Monte Carlo simulated and computed X-ray spectra with measured spectra.
Med. Phys. 26, 303–309.
simulation problem or specific additional features of both codes. Birch, R., Marshall, M., 1979. Computation of bremsstrahlung X-ray spectra and
MCNP4C has more capabilities for variance reduction calculations. comparison with spectra measured with a Ge(Li) detector. Phys. Med. Biol. 24,
505–517.
For example FLUKA has not point detector tally, which is more Blough, M.M., Waggener, R.G., Payne, W.H., Terry, J.A., 1998. Calculated mammo-
efficient for calculation of X-ray spectra in various angles. On the graphic spectra confirmed with attenuation curves for molybdenum, rhodium,
other hand, FLUKA gives more accurate results for specific X-rays and tungsten targets. Med. Phys. 25, 1605–1612.
Boone, J.M., 1988. The three parameter equivalent spectra as an index of beam
which are crucial in mammography imaging. For specific
quality. Med. Phys. 15, 304–310.
problems the results of both codes should be compared with Boone, J.M., Seibert, J.A., 1997. An accurate method for computer-generating
experimental data. tungsten anode X-ray spectra from 30 to 140 kV. Med. Phys. 24, 1661–1670.
 ARTICLE IN PRESS
R. Taleei, M. Shahriari / Applied Radiation and Isotopes 67 (2009) 266–271
Boone, J.M., Fewell, T.R., Jennings, R.J., 1997. Molybdenum, rhodium, and tungsten
anode spectral models using interpolating polynomials with application to
mammography. Med. Phys. 24, 1863–1874.
Briesmeister, J.F., 2000. MCNP—a general Monte Carlo N-particle transport code.
Report LA-13709-M, Version 4C Los Alamos National Laboratory, Los Alamos,
NM.
Fasso, A., Ferrari, A., Roesler, S., Sala, P.R., Battistoni, G., Cerutti, F., Gadioli, E.,
Garzelli, M.V., Ballarini, F., Ottolenghi, A., Empl, A., Ranft, J., 2003. The physics
models of FLUKA: status and recent developments. In: Computing in High
Energy and Nuclear Physics 2003 Conference, paper MOMT005.
Ferrari, A., Sala, P.R., Guaraldi, R., Padoani, F., 1991. An improved multiple scattering
model for charged particle transport. Presented at the International Conference
on Radiation Physics, Dubrovnik.
Fewell, T.R., Shuping, R.E., 1978. Handbook of Mammography Spectra. HEW
Publication (FDA), 79–8071.
Fewell, T.R., Shuping, R.E., Healy, K.E., 1981. Handbook of Computed Tomography X-
ray Spectra (HHS Publication) (FDA) 81-8162. US Govt. Printing Office,
Washington, DC.
Halbleib, J.A., Kensek, R.P., Mehlhorn, T.A., Valdez, G.D., Seltzer, S.M., Berger, M.J.,
1992. ITS Version 3.0: Integrated TIGER Series of Coupled Electron/Photon
Monte Carlo Transport Codes, SAND91-1634.
Hughes, H.G., 1997. Status of electron transport in MCNP. LA-UR-97-1368. Los
Alamos National Laboratory, Los Alamos, NM.
Kulkarni, R.N., Supe, S.J., 1984. Monte Carlo calculations of mammographic X-ray
spectra. Phys. Med. Biol. 29, 185–190.
Landau, L.D., Pomeranchuk, I.Ya., 1953. The limits of applicability of the theory of
bremsstrahlung by electrons and of creation of pairs at large energies. Dokl.
Akad. Nauk SSSR 92, 535 (in Russian).
Landau, L.D., Pomeranchuk, I.Ya., 1965. Electron-Cascade Processes at Ultra-high
Energies, Collected Papers of L.D. Landau. Pergamon Press, Oxford,
pp. 589–593.
271
Mercier, J.R., et al., 2000. Modification and benchmarking of MCNP for low-energy
tungsten spectra. Med. Phys. 27, 2680–2687.
Migdal, A.B., 1956. Bremsstrahlung and pair production in condensed media at
high energies. Phys. Rev. 103, 1811–1820.
Migdal, A.B., 1957. Bremsstrahlung and pair production at high energies in condensed
media. Zh. Exp. Teor. Fiz. SSSR 32, 633–646 Sov. Phys. JETP 5, 527–536.
Ng, K.P., Kwok, C.S., Tang, F.H., 2000. Monte Carlo simulation of X-ray spectra in
mammography. Phys. Med. Biol. 45, 1309–1318.
O’Meara, J.M., Chettle, D.R., McNeill, F.E., Prestwich, W.V., Svensson, C.E., 1998.
Monte Carlo simulation of source-excited in vivo X-ray fluorescence measure-
ments of heavy metals. Phys. Med. Biol. 43, 1413–1428.
Sanborg, M., Dance, D.R., Alm Carlsson, G., Persliden, J., 1994. A Monte Carlo for the
simulation of image quality and absorbed dose in diagnostic radiology.
Comput. Methods Prog. Biomed. 42, 167–180.
Seltzer, S.M., Berger, M.J., 1985. Bremsstrahlung spectra from electron interactions
with screened nuclei and orbital electrons. Nucl. Instr. Methods B 12, 95–134.
Seltzer, M., Berger, M.J., 1986. Bremsstrahlung spectra from electrons with kinetic
energy 1 keV–10 GeV incident on screened nuclei and orbital electrons of
neutral atoms with Z ¼ 1–100 At. Data Nucl. Data Tab. 35, 345–418.
Ter-Mikaelyan, M.L., 1954. Bremsstrahlung radiation spectrum in a medium. Dokl.
Akad. Nauk SSSR 94, 1033.
Tucker, D.M., Barnes, G.T., Chakraborty, D.P., 1991. Semiempirical model for
generating tungsten target X-ray spectra. Med. Phys. 18, 211–218.
Verhaegen, F., Castellano, I.A., 2002. Microdosimetric characterisation of 28 kVp
Mo/Mo, Rh/Rh, Rh/Al, W/Rh and Mo/Rh mammography X-ray spectra. Radiat.
Prot. Dosim. 99, 393–396.
Verhaegen, F., Nahum, A.E., Van de Putte, S., Namito, Y., 1999. Monte Carlo
modelling of radiotherapy kV X-ray units. Phys. Med. Biol. 44, 1767–1789.
Wilkinson, L.E., Johnston, P.N., Heggie, J.C., 2001. A comparison of mammography
spectral measurements with spectra produced using several different
mathematical models. Phys. Med. Biol. 46, 1575–1589.