Dosimetric Consequences of Dropping the Momentum Analysis
Norwegian University of Science and Technology
TFY4500 Project Thesis
Dosimetric Consequences of
Dropping the Momentum Analysis System
in a Compact Proton Therapy System
Pål Erik Goa
December 20, 2014
The high investment cost is together with the area footprint one of the great hinders of proton therapy.
The introduction of compact proton therapy systems sets out to lower both of these. But one of the
newly adjustments, the dropping of the momentum analysis system, might have clinical consequences.
The momentum analysis system normally alters the energy spread of the clinical proton beam, and hence
the shape of the Bragg peak and the distal dose falloff.
This possible difference of the distal dose fall have been quantified for all treatment ranges. Fluka, a
Monte Carlo based software was used to simulate different beam setups by dropping the proton beam in
a water phantom. The distal dose falloff were read out and compared to the distal dose falloff of systems
with an momentum analysis systems.
The results have shown that the distal dose falloff is constant for a compact proton therapy system
without an momentum analysis system. One setup gives a falloff of 5.88 ± 0.08 mm , an other gives a
shorter falloff of 5.15 ± 0.08 mm. These values are realistic values, compared to similar measurements of
a existing compact proton therapy centre. The falloff depends on the parameters of the extraction beam
from the cyclotron, and inherit the distal falloff of the maximum treatment range.
Compared to standard proton therapy system, there is no clinical consequences for deep seated
tumours, but there might be a difference in the dose to critical organs for shallow tumours, such as
breast cancer. This has been further discussed, but the measurements are to plain to make any clear
conclusions. It is recommended that this potential dosimetric consequence is investigated further for
clinics who is interesting in obtaining such an compact proton therapy system.
I would like to thank everyone who have supported me during the work with this thesis. Since I started
to look for the possibility to write a thesis on proton therapy, I’ve only been met by support and good
will from all parts. I has been a motivating semester.
First I would like to thank my supervisor, Pål Erik Goa, whom “hijacked” me and gave me the
opportunity to write a proton therapy thesis at my own home university. I’ve appreciated that he
arranged weekly meetings, even though he must have the tightest schedule at the whole university. I
have enjoyed the advices, the professional discussions and the relaxed atmosphere.
I would also like to thank Odd Harald Odland of Haukeland University Hospital and Dieter Röhrich
of University of Bergen, whom openly replied to my request on writing a thesis. I’m grateful for the
cooperation and all the help given. My simulation work relies upon the knowledge I gained at the
workshop arranged by Dieters group. I would also thank University of Bergen for the personal funding
of attending this workshop.
From this group, I would specially thank Kristian Ytre-Hauge for helping me out with the simulations,
both with the technical aspects and the discussion of parameter setting.
I also appreciated the interest shown from Jomar Frengen, Sigrun Saur Almberg and the others at
the radiotherapy clinic of St. Olav University Hospital.
I really glad the John Alfred Brennsæter joined me on writing a thesis on proton therapy. It has
been great to have someone locally to ask about any technical difficulties and discuss results.
I’m grateful for the everlasting support from my parents, including the financial support to buy a
new powerful computer. I would not have been able to write the thesis without that one.
BP . . . . . . . . . . . . . . Bragg peak
CNS . . . . . . . . . . . . . Central Nerve System
CTV . . . . . . . . . . . . Clinical Target Volume
DDF . . . . . . . . . . . . Distal Dose Falloff
ESS . . . . . . . . . . . . . . Energy Selection System
GTV . . . . . . . . . . . . Gross Tumour Volume
HCL . . . . . . . . . . . . . Harvard Cyclotron Laboratory
ICRU . . . . . . . . . . . International Commission on Radiation Units & Measurements
IGRT . . . . . . . . . . . . Image Guided Radiation Therapy
IMPT . . . . . . . . . . . Intensity Modulated Proton Therapy
IMXT . . . . . . . . . . . Intensity modulated x-ray therapy
ITV . . . . . . . . . . . . . Internal Target Volume
LET . . . . . . . . . . . . . Linear Energy Transfer
linac . . . . . . . . . . . . . Linear Accelerator
MAS . . . . . . . . . . . . Momentum Analysis System
MGH . . . . . . . . . . . . Massachusetts General Hospital
MLC . . . . . . . . . . . . Multi Leaf Collimator
NTCP . . . . . . . . . . . Normal Tissue Complication Probability
OAR . . . . . . . . . . . . Organ at Risk
PSI . . . . . . . . . . . . . . Paul Scharrer Institute
PTV . . . . . . . . . . . . . Planning Treatment Volume
RBE . . . . . . . . . . . . . Relative Biological Effectiveness
SOBP . . . . . . . . . . . Spread out Bragg peak
TPS . . . . . . . . . . . . . Treatment Planning System
1.1 Historical background of Proton therapy . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Purpose of project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1 The Physics and Science of proton therapy
2.2 The Technology of Proton Therapy . . . . .
2.3 Clincal aspects of proton therapy . . . . . .
2.4 In silico methods . . . . . . . . . . . . . . .
2.5 Summary of relevant theory . . . . . . . . .
3.1 Different setups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2 Simulation and detecting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.1 Preliminary results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.2 The distal dose falloff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.1 Analysis and validity of model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
5.2 Clinical impact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
A Commercial compact proton therapy systems
Historical background of Proton therapy
In the end of 1895 Wilhelm Röntgen created and observed a new type of radiation, and he named it x-ray.
In his famous paper ”On a new kind of rays” he claimed that he could see the bones inside the body.
Physicians around the world reacted quickly upon this, and a new field in diagnostics called radiology
was created. During the imaging of patients they observed that the skin was burned by the rays, similar
to the UV-rays of the sun.. The beams could then also have therapeutic effect, and already in the end
of January 1986 the first breast cancer patient was treated with x-rays.
In the same year Henry Becquerel discovered radioactivity. Marie Curie worked together with
Becquerel in his lab, and together they found different ideal isotopes for radiotherapy. While the cathode
tubes of that time delivered X-rays with low energy, isotopes could deliver in a MV scale. This made it
possible to treat tumours deeper in the body, not only those close the skin.
Figure 1.1: Comparison of a depth dose distrubutions of photon and proton therapy. The dose
curve of photons have a short build up and a long exponential decaying tail, which make it practically
impossible to form a homogeneous dose from a single field. The depth dose curve of protons on the
other hand, have a small entrance dose and a distinctive peak, the Bragg peak, with a steep dose
falloff on the distal part. This unique ionization is utilized in proton therapy to create a homogeneous
dose within the 2% limits, by the “Spread out Bragg Peak” (SOBP). Courtesy of W.P. Levin et. al
In 1904 William Henry Bragg experimented with the newly discovered radioactive isotopes. By using
a α emitting radium isotope, he found the ionization done by the kinetic α particle on different depths
of the target. It was assumed that the curve was exponential decaying like the x-ray ionization curve.
Instead the curve had an entrance plateau with low ionization and a high peak of high ionization at the
end of the trajectory, see figure 1.1.  Today both the full curve and the peak is named after him.
In 1946 The American physicist Robert R. Wilson proposed to use fast protons in cancer therapy.
He had worked on cyclotrons in the Manhattan project during the war and had great knowledge on the
properties of high energy particles. He argumented for that the Bragg curve made an ideal depth dose
curve for the treating tumours and minimize the damage to the tissue. Robert R. Wilson hence marked
the start of proton therapy and is today seen at the ”father of proton therapy”.
Over the next couple of years particle research centres around the world started with initial experiments. The first patient was treated already in 1954 at Berkeley Radiation Laboratory. The first
European was treated three years later in Uppsala. During the three next decades other research centres
around the world starting treating patients as a part of their particles, including Russia, Japan and
Switzerland. But the Harvard Cyclotron Laboratory in Boston (HCL) was the centre most dedicated
in proton therapy. Between 1961 and 2002 they treated almost 10 000 patients, and the knowledge and
research from HCL is the scientific foundation of proton therapy as we know it today.
After 35 years of treating patients in research centres, originally made for particle physics experiments,
the first hospital based centre was established at Clatterhill in Great Britain in 1989. This is small centre
for eye treatment, based upon a low energy cyclotron. In 1990 the first full scale clinical centre was opened
at Loma Linda Hospital. It had is own designated synchrotron, made by physicists of Fermilab. This
setup, with multiple treatment rooms for one single accelerator, became the standard for most of the new
centres, optimizing the cost per patient. Since the centre does not share any beam time with other particle
experiments, Loma Linda treats patients all year around, not only for a short experimental time. The
number of patients treated per year exponentially boomed after Loma Linda and other centres started
it operations. The number of different facilities and treatment rooms have been growing exponentially
since the clinical introduction in 1989 and more than 90 000 patients have been treated, see figure 1.2.
Figure 1.2: Overview of the number of treatment rooms and the number of patients treated yearly.
Reprinted from the Norwegian report of 2013. 
Proton therapy is not principal different to photon therapy, like surgery of chemotherapy. So it won’t
seen as a supplementary treatment to photon therapy, but a competitor. But even through protons in
general are more effective than intensity modulated x-ray therapy (IMXT) or radiosurgery, it cannot
compete at cost per treatment. Proton therapy would always be more expensive compared to photon
therapy, so the questions if it is worth the extra cost. One report says the cost ratio is 2.4 today, but
could be 1.7 over the next 5 to 10 years. But proton therapy could still be profitable, due to the
reduction of secondary tumours and other side effects. There are different economical models, where the
main factor on the age of the patient. While proton therapy could be very effective for children, e.g.
with brain melanoma, the gain to cost ratio is still low for seniors, e.g. prostate cancer.  The
cost effectiveness would be one of the major focus areas of the community in the coming years. Proton
therapy still have to prove that it is worth the money spent.
As a response to the high investment cost, the compact proton therapy centre was introduced. While
most centres was based upon multiple treatment rooms and one accelerator, different companies today
is also delivering single room centres. These are compact centres, suitable to fit in the radiation therapy
department of existing hospitals. Two companies, IBA and Mevion, have installed the first centres already
and are started treating patients last spring. There are many approaches to the cost reduction, like
building smaller accelerators, smaller gantries or cutting the area footprint. Some commercial solutions
are presented in appendix A at the end of the thesis. 
Purpose of project
One of the companies, Mevion, chose a rather unconventional, but ingenious solution for their centre.
By mounting the cyclotron on the the gantry, they where able to fit the all equipment into a single
room. And they kept the opportunity to treat from all angles and at all depths within the body. But by
implementing the cyclotron onto the gantry, there was no room for a momentum analyser in the beam
line. The momentum analyser is able to get rid of extra unwanted neutrons, as well as reduce the energy
spread of the beam. According to J.M. Schippers and A.J. Lomax, this difference in energy spread would
enlarge the distal dose falloff of the SOBP, which again would yield an extra unnecessary dose to other
The extra organ dose is difficult to measure, and changes from patient to patient. Instead, I have
rather focused on finding the distal dose falloff of the SOBP, or just the distal dose falloff of the last
pristine Bragg peak to be correct. The distal dose falloff of a pristine bragg peak is a generic and
physical distance, independent of which supplier of the momentum analyser free system. The parameter
was quantified for different treatment depths within the body by simulating a proton beam with a Monte
By quantifying the distal dose falloff as a function of treatment depths, it is be possible to compare
the system with a momentum analyser to the systems without a momentum analyser. Further, it is
possible to rule out if the difference between them are neglectable in a clinical reality or not.
The distal dose falloff of a pristine peak has earlier been measured with a ionization chamber in
a water phantom at the first Mevion centre. According to C. Bloch et al., the distal dose falloff was
constant around 6 mm for all depths.. The constant falloff of the Mevion centre has been confirmed
by another paper from the same centre, where the distal dose falloff of the full SOBP was reported to
be constant around 7mm.
The Physics and Science of proton therapy
Bragg also described the stopping power of different target materials, the retarding force on the particle
inflicted by the matter, mainly by the electrons. Niels Bohr introduced in 1913 a theoretical model
for the stopping power, and defined the stopping power to be the ”linear rate of energy loss per unit path
length”: S = − dE
dl . The most common units for S is KeV/µm or MeV/nm. Bohr deduced his model
upon classical physics and was a bit low compared to the experimental values.
But in 1930 Hans Bethe analytically deduced a new model based upon quantum mechanics. After
some corrections and including relativistic effects, the Bethe formula became known as:
4π nz 2
me c2 β 2
1 2me c2 β 2 γ 2 Tmax
− β2 −
where n is the electron density of the medium, e is the elementary electric charge, 0 is the permittivity
of vacuum, Tmax is maximum kinetic energy of the particle and δ(βγ) is a density correction. The bethe
formula gives the complete function for the the energy loss, and includes both the force from nucleus
and the electrons of the target material.
In many cases it is also neat the general stopping power of a particle, independently of the density ρ
of the target material. This known as mass stopping power :
which in the convenient units
When the light protons come to close to the heavy nuclei of the target material, the protons will have a
angular shift in the trajectory due to Coulomb scattering. Since the material is packed with nuclei, many
such events would happen. The angular shift will then have a gaussian statistical distribution, with the
mean at 0 degrees. The sigma of the distribution is then dependent on the matter which the protons are
passing through. Light nuclei with low Z, like helium or carbon, will give an low sigma, while high Z,
like lead, will give an high sigma. Another important difference between proton beam and the photon
or electron beam is the level of scattering. Since protons are much heavier compared to electrons, they
also scatter less in every interaction. This means that the penumbra is much smaller, giving a higher
precision for proton beams. In the same way, other heavier ions will scatter even less than protons.
The last interactions is nucleus-nucleus interactions, where the accelerated protons enters the nuclei of
the matter and excites them. The protons are then able to knock out other particles, like protons,
neutrons, alpha particles or other ions. Some protons would just stay put, and a create new elements.
Some of the new elements could be radioactive isotopes, which will collapse at some time later. While
the stopping and scattering of the protons is only able to absorb a small fraction of the total energy,
nuclear interactions could transform all of the energy. 
As the protons are stopped and slowed down, most of the energy is transferred by ionizing the atoms
in the tissue locally. The absorbed dose is then defined as the ionization done by the energy from the
protons per mass of tissue:
Dose is today measured in Gray (Gy), where 1Gy = Kg
. If dN numbers of protons is passing through
an infinitesimal cylinder of cross sectional area dA, thickness dl, we have
−(dE/dl) × dl × dN
ρ × dA × dl
dA is the infinitesimal proton fluence, the numbers of protons passing through an area. Fluence
is mathematically defined as
The absorbed dose from the protons can then be written as
This means that the absorbed dose D is dependent upon the fluence Φ, which if also shown in figure
The energy/range relation of Bragg curve
The Bragg curve, which is the relative depth dose curve, gets its characteristic shape from these three
main interactions which occur between the accelerated protons and matter. The shape of the ionization
curve is strongly dependent on the energy of the protons. Rangestraggling, a phenomenon described
in section 2.1.6, is one of the energy dependent shape parameters. The range which the protons travel
into matter before they come to a halt, is another. The mean range R0 can be calculated using the
“continuous slowing down approximation“, by integrating over the stopping power for the maximum
energy E0 to 0:
R(E0 ) = R0 =
where the p coefficient is dependent of the particle type and α is dependent of the absorbing medium.
International Commission on Radiation Units & Measurements (ICRU) have experimentally given α ≈
2.2 × 10−3 cmMeVp and p ≈ 1.77 for protons passing through water, as shown in figure 2.2. The mean
range R0 the particles travel is approximately around d80, at the distal side of the Bragg peak, where
the relative dose is 80% of the total dose. This is also shown in figure 2.4. 
Figure 2.1: Measured data of seven Bragg curves in water with different energies at PSI, Switzerland.
The solid lines represent the curve done by the model calculation. The range straggling phenomenon,
described in section 2.1.6 can be seen by the different shape of the Bragg peak at different ranges. At
97 Mev the peak is narrow, at 214 Mev the peak is wide.
Figure 2.2: The relation between mean range and the energy of accelerated protons in water. The
errorbars are measurements are taken from the ICRU 49 report, the line fitted to equation 2.8. Made
by T. Bortfeld 
Since all of the particles interactions (stopping, scattering and nuclear interactions) are randomized
processes, the particles in a beam will not have the same number of collisions. They are stopped and
scattered differently, so every proton would not have the same endpoint (range) in the material, even
for monoenergic protons in a homogeneous material. The endpoint of the different protons are normal
distributed, with σR . This statistical phenomenon is called range straggling. The range straggling in
a monoenergic beam is higher for particles with high energy compared to particles with low energy, as
seen in figure 2.3, which is almost proportional. According to an approximation by T. Bortfeld, the
σR = σR,mono = 0.012R00.935
for protons in water.
Figure 2.3: Straggling as function of range. Comparison of a proton, helium and neon beam in
water. Range straggling corresponds to the standard deviation σR in Bortfelds equation 2.9 of the
range distribution of a monoenergic beam with mean range R0 . A good approximation of the straggling
varies inversely to the square root of the particles mass. 
But figure 2.3 and equation 2.9 hold for ideal monoenergical beams, and it does not give the full
picture of range straggling and the shape of the Bragg peak caused by the energy spectrum. The energy
spectrum of clinical proton beams does not have the ideal monoenergical δ-shape. The energy spectrum
are normally normal distributed around and energy E with an σE . This will also affect the endpoint of
the protons and the σR . But Bortfeld also made an approximation that include the energy spectra. If
the beam enters the water phantom with an energy spread σE,0 around E0 , and σE,0 << E0 , the new
estimate would be:
α2 p2 E02p−2
The approximation of the variance of the protons endpoint is then a function of either the range R0
or the initialized E0 of the protons. The other components are known for protons in water, besides σE,0 ,
which is dependent on the proton therapy system.
And as shown in figure 2.4, changing σR will also change the shape of the Bragg peak. A small σR
will make a steep Bragg peak, a big σR will smear out the Bragg peak. Later, in subsection 2.3.2, a
clinical method of defining the dose falloff at the distal part of the peak is introduced, DDF . As the
shape of the Bragg peak is dependent of the σR , this parameter should then also be dependent upon σR .
In general we have
Figure 2.4: Comparison between the proton fluence (a), the endpoint or range (b) and the corresponding depth dose curve (c) of three different beams. The beams have the same mean range R0 ,
but the σR is different. The gaussian curve defined by the σR is shown in the (b). The smearing of
the Bragg peak shape (larger DDF ), due to a higher σR is shown between (b) and (c). It should also
be noted that the mean range R0 correspond to d80, as shown in (c). This is not the same as the
clinical range R defined in section 2.3.2, where it is defined as d90. Courtesy of B. Gottschalk. 
Biological properties of a proton beam
Most of the particle’s energy that is lost due to the stopping power is transferred to the target material,
and deposited locally at the same place. A small portion of the energy is converted to bremsstrahlung
or long range delta electrons, so the energy is transferred to another place in the material. But this is
only a small fraction of the total energy in the case of protons. It is therefore possible to simplify the
linear energy transfer (LET) to:
LET is a parameter that defines the physical ionization in the material that is radiated. Protons
are categorized as low LET, together with x-rays and γ-rays. Low LET particles ionizes more sparsely
than the densely ionizating high LET particles. So in the linear quadratic radiobiology model, low
LET particles have in principle higher α/β ratio compared to high LET. But LET does not give the
full picture of the biological dose, it only dense the particles are ionizing. Hence another parameter
is more important, namely the relative biological effectiveness (RBE). RBE is based on experimental
measurements, where the same absorbed physical dose will induce a higher number of cell deaths for
protons (in our case) relatively to the same dose from 250 kv x-ray photons. RBE is also strongly
dependent on which kind of tumour tissue. It have also been shown that the RBE value have a small
rise at more distal parts of the Spread Out Bragg Peak (SOBP, explained at subsection 2.3.2), from a
ratio of 1.0 to all the way up to 1.7, as seen in figure 2.5.
Figure 2.5: The physical dose and biological dose simulated for one particular endpoint and dose.
The biological is the product dose between the physical dose times the depth depending RBE. As
seen in this figure, the RBE is not perfectly constant 1.1, but has a small peak towards the end of the
SOBP. Reprinted from H. Paganetti 
But RBE is not a simple value to measure or calculate, so this variation is actually neglected in
the clinics today. As a great simplification, the RBE value of protons is normally set to 1.1. So by
“neglecting” the variation in the RBE, a clinical proton beam creates in principal the same effect and
damage to the cells as gamma and x-rays. Therefore the only main difference between a conventional
photon beam and a proton beam is the shape of the dose curves, both the lateral- and depth curves.
Since the LET and RBE is constant, the depth dose curve of protons have the same shape, range and
relative intensity as the ionizating Bragg curve, see figure 2.1.
All charged particles follow the Bethe formula, so the Bragg curve is not particular for protons. Other
particles have also been investigated the last 60 years, both their physical and their biological properties.
Protons did not only had to prove that it was better than conventional photon therapy, it also has to
be superior to other kinds of experimental radiation therapy. Fast neutrons was for a long time seen as
a more possible candidate for wide clinical usage, due to the low kinetic energy cost. Another variety
called neutron boron capture therapy used a neutron accepting compound based upon the element boron
to harm the cancer cells, working like a localized radio pharmaceutical. Minus pion was also investigated
in external particle therapy at Los Alamos and Paul Scherrer Institute (PSI) for many decades. Today
only heavy ions, specially carbon ions, is left for patient treatment. 
The Technology of Proton Therapy
While radioactive isotopes is suitable for external beam photon therapy, the maximum kinetic energy of
α radioparticles are about 5 MeV. At this energy they won’t even penetrate the skin, only a couple of
cells. So the protons have to be artificially accelerated to reach the clinical energies corresponding to
the tumour depth.
The most central force in particle acceleration is the lorentz force
F = q (E + v × B)
As a consequence of this law, which is not deduced here, it is only possible to accelerate charged
particles by an electrical potential. The magnetic field would only bend the trajectory of the particle,
not add any kinetic energy.
In 1927 the Norwegian physicist Rolf Widerøe developed the linear accelerator (linac). It uses an
electric potential to accelerate electrons or other charged particles in a cavity. Linacs is used for
accelerating electrons for electron and photon therapy, but linear accelerators for protons would be too
long to use in a clinical setting. Fortunately it is possible to gain higher kinetic energy by using circular
accelerators instead. In contrast to a linac, circular accelerators reuse the accelerator cavity to further
increase the velocity of the particles. To bend the particles, a perpendicular magnetic field to the particle
like the magnetic force component of the
trajectory is used. By setting the centripetal force Fz = mv
lorentz force, we achieve:
= qvB = Fmag
Since the charge q and rest mass m0 is physical constants for protons, either the magnetic field B
and/or trajectory radius r needs to be enlarged when the velocity v (or kinetic energy) is getting larger.
Circular accelerators is divided in two main categories, either cyclotrons or synchrotron, depending of
which of these parameters is enlarged.
Figure 2.6: Sketch over the principle of a cyclotron, consisting of two dees. The ions are accelerated in
the electric field between the dees and bended in the dees. They are then returning to the accelerating
cavity, which is then flipped. This is repeated until the beam have the right circumference and
velocity. It could there just be extracted a continuous beam with a single energy. Reprinted from
The cyclotron was invented in 1930 by E.O. Lawrence. The cyclotron consists of two dipole magnets with
a constant magnetic field B over a wide area. Charged particles placed in the middle of the cyclotron,
where they are accelerated by in a cavity with electrical potential. The particles is then bent 180 degrees
and returned to the cavity, which now have an opposite potential, which accelerates the particle further.
The particle is then bent back 180 degrees, but since the velocity is higher, the radius r is larger for
every time it passes the accelerator cavity. The electrical potential is changed by an radio frequency
pulse. If this pulse is constant, particles at different radii could be accelerated at the same time. That
means that angular velocity ω = vr needs to be constant. We could then rewrite equation 2.16
If the particles should be able to treat the deepest tumours, the particles need to be accelerated up
to a energy about 250 Mev. This corresponds to 0.6c, which mean we need to include relativistic effects.
The equation is then corrected to
where m0 is the rest mass and γ is the relativistic Lorentz factor. While the B field can be constant
for all radii for non relativistic energies, B got to be slightly stronger as r gets higher for relativistic
velocities. Such an cyclotron is called isochronous cyclotron, where the beam is then continuous, or
quasi-continuous to be correct. A sketch of a cyclotron is shown in figure 2.6. In an alternative type
of cyclotron, a synchrocyclotron, the radio pulse does also variate, so that a particles is accelerated
in bunches. The downside for a synchocyclotron is that the current and the extraction rate is lower.
Anyway, a synchrocyclotron could normally have higher B field and be made much smaller compared to
the isochronous, which saves space, material and money for the user.  The cyclotron is only able to
extract the protons at a fixed energy, which in clinical cyclotrons is set to equal the maximum treatment
range, about 230-250 MeV. A energy selection system is described in the next subsection, which able
cyclotron based centres to treat at shorter ranges as well.
Figure 2.7: Overview of the clinical synchrotron at Loma Linda just outside LA. A bunch of protons
are ionized in the ion source and then preaccelerated through the straight linac. The protons are led
into the synchrotron, where they are further accelerated by the acceleration cavity in the lower left
corner. The magnet field in the bending magnets get higher for every turn, until the protons reach
their wanted energy. The protons are then extracted from the synchrotron through the beam line to
the upper right, towards the treatment rooms. Reprinted from .
In 1944-1945 the synchrotron was independently invented in the USA and Soviet Union. It also reuses
an electric potential to accelerate the particles as the cyclotron. But instead of increasing the radii of
the beam r, it is increasing the magnetic field B of the bending magnets for every turn, as in figure 2.7.
The packages of particles, called bunches, would be extracted from the closed beam line when they have
reached the wanted energy.  
Energy selection system
Cyclotrons are only able to extract their beam at the maximum treatment energy, so the beam has to
be degraded to achieve lower energies. This is done by sending the beam through a low Z material like
PMMA, which slows down the beam and loses some of its energy. Such a beam component is normally
called a degrader or rangeshifter, and normally placed directly after the cyclotron, as shown in the IBA
system in figure 2.8. The degrader is able to change the energy by adjusting the thickness of the material
which the beam has to pass through. This could by done in many ways. Some system uses a water filled
piston, some adjustable wedges uses a rotatable disk with degrader. The energy adjustment should be
done as fast as possible to get rid of interpolation effects, one of the greatest issues in modern proton
Figure 2.8: A drawing of the ESS used in IBA Proteus PLUS. The beam is extracted from the
cyclotron to the left at 230 MeV, before the Energy moderator (degrader) degrades the energy E of
the beam. The lateral scatter from the degrader is filtered away by the transverse collimator. The
beam is bent by the dipole magnets, so that the momentum collimator is able to filter away the lowest
and highest energy, and lower the σE . The Reprinted from Kostromin et al. .
Momentum Analysis System
Unfortunate, the straggling phenomenon also is present in the degrader, so some particles are slowing
more down than others. In practice this means that the energy E of the beam is not just degraded, the
energy spread σE of the beam is enlarged. And as mentioned in equation 2.10, the σR is dependent of
σE . So to get a smaller σR , the σE should be made smaller. This is done by bending the beam through
a turn with the help of dipolar magnets. Equation 2.16 could then be rewritten to
or as it is in the relativistic form
where γm0 v is the relativistic momentum p. If the magnetic field B is then static and uniform, the
radii of the protons will be linear proportional to the energy of the protons. A collimator placed in the
beam line is then able get rid of the protons with low radii (low energy) and those with a high radii
(high energy). This collimator, a beam component called “momentum slit”, is variable such that the
wanted σE of the beam could be chosen. The upper limit is determined by the engineering done by the
supplier, and is typically about 1%. The lower limit is normally limited by the dose rate, since a
thin momentum slit would slice away a lot of protons, and drastically lower the dose rate. The bending
magnets and the momentum slit are the components of the the momentum analysis system (MAS).
Together with the degrader make up the energy selection system (ESS), as fully shown in figure 2.8.
The MAS has also important function in filtering away neutral neutrons produced in the degrader.
But this topic is not discussed in this thesis, but these could also give an extra unwanted dose to the
Treatment room and gantry
Along with Loma Linda came many new great developments in the field of proton therapy. The treatment
room where optimized for treating patients. Unlike the research centres, where the patients had to adjust
to the research environment. Robotic treatment tables or chairs were installed in every room for patient
positioning. Loma Linda also introduced a gantry based beam line. By using a gantry the patient could
be radiated from every possible angle and still be lying down. This is the same degree of freedom as
photon therapy, and an important improvement for outclassing traditional radiation therapy. The beam
eventually ends up in the “nozzle”, the last straight part of the beam line before the beam hits the
patient. The nozzle is in charge of shaping the beam, analogue to the treatment head of conventional
The Spread out Bragg Peak
When Robert R. Wilson proposed the use of protons in cancer therapy in 1946, he also described that by
modulating the energy of the proton beam, it is possible to produce an uniform dose to the whole depth of
the tumour. Uniform dose to full tumour volume is an important principle of radiation oncology. While
conventional radiation therapy normally use stereo-tactic techniques to create a uniform dose, protons
are able to fulfil the principle from one single field. The new composite depth dose curve adapted to the
individual tumour size was later named ”Spread Out Bragg Peak” (SOBP), as seen in figure 2.9.
Figure 2.9: Plot over a constructed SOBP with multiple Bragg peaks with different energies. The
SOBP in red curve is the sum of the 13 individual Bragg peaks, as seen in black with blue dots.
The dosimetric range, the position of the distal 90% dose, is marked by a red dot at 25.0 mm. The
modulation, the depth of the SOBP is 10.0 mm, which is the distance to the second red dot at 15.0
mm. As this example of a SOBP illustrates, the distal dose falloff of the SOBP is made mainly out
from the shape of the most distal principal Bragg peak. In this example the DDFBP is 0.50 mm,
while the DDFSOBP is 0.53 mm, slightly higher (ρ = 1.06). The plot is made by a python script of
L. Grzanka, IFJ PAN Cracow 
With a finite and limited number of individual Bragg peaks, it is impossible to get a complete flat
SOBP. The plateau of the SOBP will always be slightly oscillating, as it is a composite of a finite
number of gaussian shaped Bragg peaks. Different inverse optimizing algorithms have been made, with
different number of individiual peaks, different weighting of the individual peaks and without an isoenergic distance. In all cases, the last Bragg peak is always the major one. But there is a give and take
of the different algorithms. By weighting the last Bragg curve high, such that the distal falloff of the
SOBP is only defined by the distal falloff, the SOBP will be highly oscillating. To minimalize the this,
the last Bragg peak is therefore often weighted less than 100% of the maximum dose, as seen in figure
2.9. The distal falloff of the SOBP is then a bit higher than the distal falloff of the most distal Bragg
peak. Later, in subsection 2.3.2, a clinical definition of the distal falloff is introduced, equation 2.23.
By using this, we could mathematically describe this correlation:
DDFSOBP = ρDDFBP
where DDFSOBP is the distal falloff of the full SOBP, while the DDFBP is only the falloff for the last
one. ρ is dependent on the SOBP algoritm used, as must be ≥ 1. This gives us
DDFSOBP ≥ DDFBP
According to Wilson such a depth dose curve would be technical achievable by placing a rotating wheel
with variable thickness in the beam between the source and the patient. The wheel will then distribute
the dose to the tumour along the axial direction. The beam should also be scattered, either once or
twice, to create a broad beam, covering more than the full lateral projected areal of the tumour. A
personalized collimator made by a high Z material would then shape the beam to the projected areal of
the beam eye. Some passive scattering systems also add a compensator for shaping the distal part of the
SOBP to the tumour shape, see figure 2.10.
Figure 2.10: Sketch over a double scattering system, with every component. The black line to the
left is the first scatterer, a thin foil, normally made of thin lead. To the right of that is the range
discriminator and the range modulator shown as a ”staircase”. A second scatterer is in this setup used
to give a flat beam profile for the whole tumour volume. After that cone shaped shielding is applied
to reduce the radiation to the patient. The last equipment in the beam shaping rack is the collimator
with the compensator. The collimator forms the beam in lateral dimensions, while the compensator
shapes dose to fit to the distal part of the tumour. The result, both in axial and lateral dimensions
are given to the right. The sketch is not to scale. Reprinted from B. Gottschalk. 
PSI in Switzerland and HIMAC in Japan introduced another method of field shaping in the mid 90’s,
named active scanning. Opposite to passive scattering, active scanning is not deforming a broad beam,
but instead painting the full volume with a thin beam. The thin “pencil-beam” is drawn by two dipole
magnets over the full areal of one depth layer of the tumour, pixel by pixel. This is done step wise, from
the most distal energy layer to the nearest layer, as illustrated in figure 2.11. At the end the tumour
should have a nearly uniform dose, while the healthy tissue is spared. In other words, active scanning
is the only method to obtain true intensity modulated proton therapy (IMPT), the proton equivalent to
intensity modulated X-ray therapy (IMXT). A good analogy of comparing active scanning and passive
scattering in proton therapy would be the use of multi leaf collimator and compensator in photon therapy.
The modern techniques of both are quicker, does not need any patient specific hardware in the pathway
of the beam and creates less unwanted neutron dose. 
Figure 2.11: Simple sketch describing an active scanning system. The thin beam is independently
bended by computerized electromagnets in both lateral dimensions. The beam could therefore be
spread over the whole tumour volume, step by step: pixel by pixel, energy layer by energy layer.
Reprinted from the doctoral dissertation of D.S. Parcerisa.
During the 70’s medical imaging was radically developed. The introduction of CT and MRI scanners in
clinics made a big impact in radiation therapy. In the middle of the 80’s Michael Goitein, one of the
main physicist working at HCL, invented a 3D treatment planning system (TPS) for radiation therapy.
He also introduced the term ”beam eye view”, enabling the user to see a 2D projection for the tumour,
the basis of the collimator design. By including the electron density from a CT scan, it is possible to
calculate the dose distribution of a proton beam in all dimensions quite accurate.
Imaging has also later been implemented in patient positioning and monitoring of the delivered dose,
a term known as image guided radiation therapy (IGRT). Imaging equipment is now a natural element of
every modern treatment room. While IGRT is not seen as obligatory in front of every treatment fraction
for x-ray therapy, the need for it is far higher for proton therapy. The rapid drop in the dose curves
makes it very important to give the dose perfect according to the plan, or else the whole tumour won’t
get the full dose, and fresh tissue would get a massive extra dose. 
Clincal aspects of proton therapy
Treatment planning definitions
Before a patient is treated, a treatment plan is made in the TPS. Based upon biopsy, CT-images or other
knowledge, the radiation oncologist is in charge of defining the 3D treatment volumes. The tumour should
then be localized within the gross tumour volume (GTV), including surrounding microscopic tumour
within the clinical target volume (CTV), which should maximum be able to move within internal target
volume (ITV). All of these volumes are shown in 2.12. The planning treatment volume (PTV) is then
defined around the ITV, which should be given a homogeneous dose, prescribed by the oncologist. The
dosimetrist is then in charge of conforming the ionizating radiation to the PTV such that other organs
at risk (OAR) and other healthy tissue is given as little dose as possible, or at least within the dosimetric
Figure 2.12: Graphical representation of the target volumes of interest, as defined in ICRU Reports
No. 50 and 62. Reprinted from the IAEA handbook on Radiation Oncology Physics. 
Single field depth dosimetry
Figure 2.13: A SOBP with dosimetric definitions, as used at Massachusetts General Hospital. One
dimensional projections of possible PTV and OAR is added to the bottom of the figure. p is short for
proximal, and d is short for distal. The numbers states the percentage of the dose given at this spatial
point. The range, which is the distance from the entrance to d90, is determined by the most distal
point of the PTV. The modulation (Mod98), which at MGH is the distance between p98 and d90, is
determined by the depth of the projected PTV, such that the whole tumour is given a homogeneous.
The distal falloff, here named distal margin, is at MGH defined as the distance between d20 and d80,
as shown. In this example an OAR is added behind the PTV. Some parts of this OAR would have
gained 10%-30% of the dose, if this was an actual treatment. Based upon an figure by H.M. Lu and
As described in 2.2.4, in proton therapy it is possible to achieve a homogeneous dose (with small oscillations) to the full PTV, by only using a single field. Since the Bragg curve has a steep distal dose falloff
(DDF), this is specially neat when there is an OAR behind the PTV, as shown in figure 2.13. But the
OAR might anyway get a small percentage of the dose, depending of how steep the distal falloff is. So
it would be useful to know the “steepness”. At MGH the DDF is measured as the distance between
positions where the dose is 20% and 80%:
DDF ≡ d20 − d80
Other centres use the distance between 10% and 90%, while others also measure the partial falloff
between 50% - 90% or 10%-90%. Everything is eligible, but I have chosen to stick to the MGH definition
for the distal falloff. It is also neat to use, since the 20% - 80% dose distance is standard definition used
when the lateral penumbra of photon beams is measured.
As mentioned, a projection of a possible OAR is present behind the SOBP in figure 2.13. The dose
given to it is depending on the “steepness” of the falloff the DDF. So by minimizing the DDF, the dose
to the OAR will also be minimized. 
Since the radiation damage done by a proton beam is similar to photons, all cancer patients that are
selected for photon therapy could in principle be treated by protons instead. Protons would only be
preferable to photons when the dose distribution from the proton plan would be by far superior to the
best optimal photon plan. When the tumour is close to a radiosensitive and critical organ, the photon
plan could yield an unacceptable amount of dose to the nearby OAR. A better dose conformation to
the PTV would then lower the chance that the radiation does not impact other complications. Or the
opposite, a higher dose to the nearby critical organ would enlarge the Normal Tissue Complications
When protons where clinical introduced at Loma Linda about 25 years ago, the photon treatment was
quite simple, compared to today’s cutting edge standard. In the beginning of the 90’s, many clinics in
the western world only used cobolt-60 gamma machines, which have a poor depth dose curve compared
to the modern MeV linacs. The treatment planning and treatment techniques were also quite simple.
3D Conformal Radiation Therapy (3D-CRT) was the most modern at that time, before IMXT was
introduced some years later. So at this time, the difference in the dose distribution between photon
plans and proton plans was enormous. But a lot have happened since then in the field of photon therapy.
Inverse planning, Volume Modulated Arc Therapy (VMAT) and Flattening Filter Free (FFF) is some
of the newly clinical adapted techniques which have further optimized the dose distribution for photon
treatment. According to the Norwegian report on proton therapy of 2013, about 10-15% of the patients
receiving external radiation therapy would benefit from receiving proton therapy instead. It would
be especially beneficial for treatment of some rare pediatric patients with tumours in the Central Nerve
System (CNS), eye tumours and some aggressive brain tumours, where the OARs is close by to the PTV.
The different kinds of tumours is summarized in table 2.1.
Table 2.1: The table gives an overview of the different group of cancer patients that could benefit
from proton therapy in Norway every year, in controlled studies and in standard treatment. It also
gives representation of the distal edge of the PTV for the different kind of tumours. This list is based
upon an analysis done by John Cameron and Niek Schreuder, see figure 2.14. The minimum distal
and maximum distal distances is read out for every set of category within conservative regions. The
Norwegian list of different cancer is then categorized under the most appropriate and conservative
category. Like CNS (adult) for example is categorized under brain instead of spinal cord. Some kind
of cancers, like breast, have not been analysed in the American review and the depths do not exist.
These cancers have been marked by an *, and their depths have been set by conservative depths by
the author. It should be mentioned that this depth is measured through inhomogeneous human tissue,
which in general have higher attenuation than water. Table is remade from the Norwegian Report.
Type of Cancer
Number of patients
Head and Neck
Skull Chordom/ Chondrosarcom
Other Gastrological Tumours
Bone and Soft Tissue Sarcoma
Meningioma and other Benign Tumours
Lymphoms and Thymom
Reradiation and Palliative Treatment
Children- incl. CNS
Melanoma and other eyetumours
Kidney and urinary tract
Distal Depth [cm]
Depth of different tumours
Cancer can develop in almost every kind of tissue in the body, at the skin or deep within the body.
The treatment field have need then to be adapted treat every possible kind of cancer, at any depth.
John Cameron and Niek Schreuder executed an extensive analysis over the most common and desirable
proton therapy treatment protocols in use in USA. The depth of the most distal edge of different tumour
categories are shown in figure 2.14. The values have been read out, and fitted together with the Norwegian
overview of possible patients, as shown in the table 2.1. The spinal axis is skin close, the distal part
of the PTV is at most 6-7 cm from the surface of the back. But for prostate, the distal edge of the
PTV could be 32 cm from the surface of the body, if the patients have an abdomen circumference above
average. These 32 cm is the reason why the accelerators need to be able to accelerate the protons all the
way to 230 MeV. By convolving the American analysis with the Norwegian patient data, is is possible to
get the volume (number of patients) as a function of the range needed to treat the patients, as shown in
figure 2.15. Even though 230-250 MeV is the standard extraction energy of clinical cyclotrons today, it
would still be possible to treat many patients by using smaller accelerators. As an example, 50 % of the
patients could still be treated with an maximum range of 17 cm, which according to equation 2.8 would
give an energy around 160 MeV. 
Figure 2.14: A overview over the maximum lateral and distal limits of the treatment fields of
different cancer groups. This figure is based upon an extensive analysis of actual plan or protocols
used at proton centres in USA. Work done by John Cameron and Niek Schreuder. 
Figure 2.15: The graph show the total amount of patients possible to treat as a function of the
range of the protons. The patient data is taken from the Norwegian report, while the range data
is from an American analysis. The data are also displayed in table 2.1. The number of patients in
every category where uniformly distributed for every centimetre within the minimum and maximum
range of every category, which is an conservative approximation. A normal distribution might be
better. The range data is based upon American patients, which might have larger spatial dimensions
compared to Norwegians. In reality, the graph might look a bit better, and shifted towards left, to
shallower depths. 
While the number of proton centres around the world is expanding in a high rate, proton therapy has not
yet been properly medical tested. As seen in table 2.1, most of the patients groups have not been properly
studied. There has not yet been conducted a single randomized III trial, which normally is obligatory for
every clinical treatment before it reaches the market. This has been and still is a great debate between
physicists, physicians and politicians. Because of our knowledge of the physics, radiobiology and the
current minor trials of proton therapy, proton should without doubt be better than today’s treatment.
But since medicine is based upon empirical based knowledge, this still would have to be proved by trials.
Multiple clinical trials, randomized or not, will be conducted over the next couple of years to prove our
a priori knowledge.
In silico methods
Monte Carlo simulations
In section 2.1, all of the basic interactions between a proton beam and target matter was introduced.
Stopping, scattering and nuclear interactions are the three isolated processes which happen at micro
scale between the atoms. To calculate the range of the protons, the dose given to the patient and so on,
it would be neat to know the effect of these processes. In a observable scale, these processes happen more
or less chaotically, so it is not possible to solve the problem analytically. Observations have anyway given
us useful statistical data, so we know the probability to these processes that will happen, depending of
a given time, a given angle, a given energy, a given target nuclei, etc. By using large datasets with all
possible processes, we are then able to simulate the trajectory of a single particle and the secondary
particles which is produced. These trajectories could give us different interesting information, including
the spatial distribution of the deposited energy. A single particles trajectory will not give us any useful
information, so we are simulating many thousands or millions of different “tracks”. We continue until the
information we are interested in converge against a constant value, and the statistical uncertainty is then
given according to the central limit theorem. The number of runs could then be limited by predefining
“acceptable” error margins, typically 1-2%.
Monte Carlo simulations is a comprehensive numerical particle transport method, somewhat like a
advanced “random-walk” calculations with many different processes and probability distributions. All
sums and problems in particle tracing will converge in the end. So if used in the correct manner, no
other calculation method would give a more correct answer. Monte Carlo simulations is accepted as the
“gold standard” in dose calculation in medical physics, and other methods is calibrated by it. 
Other dose calculation models in proton therapy
The Monte Carlo technique is comprehensive and is normally to time demanding for normal clinical
use. It is also practically impossible to make a good inverse plan, only based upon full Monte Carlo
calculations. Simpler computational models have therefore been made, which incompetent the most
important parts, well able to make reasonable plans for the simplest cases. Algorithms like the Pencil
Beam, Analytical Anisotropic Algorithm and Collapsed Cone Convolution are three examples of simpler
methods to calculate the dose. These are implemented into a TPS, which is then able to make fast plan
for every patient. The algorithms need to be as realistic as possible, so a general generic proton plan for
all systems would not hold. To account this, a TPS would include machine specific parameters for every
system, based upon dosimetric measurements. 
Summary of relevant theory
Equation 2.10: σR
α2 p2 E02p−2
Equation 2.22: DDFSOBP ≥ DDFBP
Dose to OAR
Dose to OAR
Figure 2.16: An overview over the pathway between the MAS and the possible clinical impact. The
different quantities of the system with MAS is compared to the similar of the system without a MAS.
The references in the middle point to the theory sections where the dependency should be notated
that this figure relies upon approximation and simplifications, like the SOBP algorithm or that only
a single field is used.
The flow diagram shows dosimetric impact of dropping the MAS, by comparing the two pathways step
by step. The pathway might have some weaknesses at some points, this only is a simplified version.
The steps from DDFSOBP and on are depending on other technology and choices, while the steps from
the momentum analyser to the DDFBP based on pure physical laws, and their limits. To get the most
general answer of the dosimetric consequences of dropping the momentum analyser, I have chosen to
only investigate DDFBP , while this is independent of any commercial solution.
For calculating the distal dose falloff as a function of range, Fluka v. 2011.2b.6 was used together with
a graphical user interface software, Flair v. 2.0-3. A simple model was made, as seen in figure 3.1. The
model was based around a small water phantom. The proton beam was dropped was i every experiment
initialized i x=0,y=0,z=-150cm and sent directly towards the water phantom. To make the experiment
as realistic as possible, all of the parameters used was gather from other research papers. The experiment
was done by comparing 5 different main setups, which focused on the different parameters from different
papers. The five setups was categorized into two categories: “The Synchrotron/ESS setups“, and the
”Degrader setups“, where the latter lacks the MAS.
Figure 3.1: Projection in the xz plane at y=0 of the model used in the in Monte Carlo Simulation.
The beam starts at z=-150cm to the left and goes straight along the z-axis. This specific figure
shows the collimator setup with a 27 cm thick degrader in yellow, the collimator in black and the
water phantom in dark blue. For the degrader setups, only the left side of the degrader changes.
The collimator is not present for the other two degrader setups. For the synchrotron/ESS setup, the
degrader is also not present. The rest is constant, besides the initial beam parameters.
Synchrotron and ESS setups
A synchrotron like beam would be the optimal reference beam, which should in theory have the smallest
distal dose falloff. The beam was initialized in z=-150 with a variable energy E0 and a beam spread
σE,0 = 0.1% of E0 . The value is taken from the patent of synchrotron at the Heidelberger IonenstrahlTherapiezentrum (HIT), a modern particle therapy synchrotron.
A system with a full ESS is able to adjust the energy spread σE,0 by adjusting the width of the
momentum collimator. The maximum spread of σE,0 is depending of the engineering done by the
company or facility. In this experiment, the maximum possible spread at the PSI facilty σE,0 = 1.2% of
E0 is used.
For both the synchrotron setup and the maximum spread ESS setup, the beam is initialized in z=150cm with a circular gaussian setup, like a common pencil beam. The parameters, σx = σy = 4.25mm,
is taken from a research paper by K. Parodi and W. Enghardt.
Since modern active scanning nozzles have no components in the trajectory of the beam (besides thin
ionization chambers), a simple synchrotron/ESS setup is easy to model. But a compact proton therapy
system would be a bit more complex, since all beam components need to be placed in the nozzle, between
the accelerator and the patient. As earlier told, there is only existing one kind of compact therapy system
without a MAS today, the Mevion S250. The passive scattering nozzle got many complex structures, as
seen in the drawing in appendix A.1.2. But the new active scanning nozzle might be made simpler (as in
number of different components in the beam). Since we are only interesting in looking at the distal dose
falloff dependency of the range, we could further simplify the model to only include the rangeshifter.
The rangeshifter is most certainly made of a material with low Z, which relatively stops the beam more
than it scatters, compared to high Z material. In the model the degrader is then set to PMMA, a typical
material in a degrader.
For simplicity, all of the setups based upon the system without a MAS, is nicked “Degrader Setup”,
due to the degrader is present in the model. In the synchrotron/ESS setups, the initial energy E0 is
varied to change the range, while in the degrader setup, the thickness of the rangeshifter is varied. In the
model the degrader has a constant endpoint at 1 meter in front of the water phantom, with a variable
thickness zdegrader from 0 cm to 30 cm i negative z direction. The degrader is the yellow component in
To mimic a realistic beam of a system without a MAS, the initial beam parameters is taken from
research papers about the Mevion system, the beam is initialized with an energy E = 250MeV in
positive z -direction at z=-150cm. In a research paper the energy spread from the synchrocyclotron is
σE = 0.42MeV.. This is then the first degrader setup. The same paper also includes the lateral
spread of the pure beam: σx = σy = 2.5mm.
The protons also scatterings a lot in the degrader, which will widen the focus of the beam at the
phantom entrance, this can be seen in the figure 4.1. Most certainly, a real setup needs have better better
focus. This could either be done by a multiple of quadrupole magnets or a single passive collimator.
There might not be any room for quadrupoles in the small nozzle, so a small passive collimator of lead
with a circular hole is chosen. In a paper about the passive scattering system in Mevion, a passive high
Z collimator is placed 2 cm in front of the water phantom. This is adapted in this setup, where the
thickness is set to a constant of 8 cm to be assure there will be no radiation behind, even though such
a piece of metal is far to heavy for clinical use. The diameter of the hole is set to 13 mm after a rough
fitting by the author. This fitting is shown in figure 4.1. This is the second degrader setup, otherwise it
is the same parameters as for the first. The collimator could be seen as the black component in figure
In the conference proceeding by Bloch et. al., the σE value from the accelerator was chosen to
1.4MeV, based upon fitting of σE to actual measurements on the first Mevion S250 in clinical use. This
value is then used as in the third degrader setup.
Simulation and detecting
10 runs was simulated of the two synchrotron/ESS setups with different energies, at steps of ∆E0 = 20
MeV from E0,1 = 70 MeV to E0,10 = 250 MeV. 11 runs was simulated of the three degrader setups, with
a stepwise variation of the degrader thickness, in steps of ∆zdegrader = 3 cm from zdegrader,1 = 0 cm to
zdegrader,11 = 30 cm. Every run tracked 5 × 105 initial protons each, which gave an ”acceptable“ error
for most runs. The dose was detected in the water phantom along the axis by a cylindrical detector.
The dose was scored for every 0.005 cm around the distal part of the Bragg peak. The small steps in
z-direction was compensated by a large radii of 4 cm. A Matlab script normalized the curve and found
the range (distal 90%) and the distal penumbra (80% - 20%) with error margins for every individual run.
The result was then plotted and the points were fitted to sensible polynomials.
To exemplify the different distal dose falloff of the different setups, two full Bragg curves for all setups
was detected, one with short range and other with a long range. Also a lateral map of the y-axis at the
surface was detected to show the the lateral spread, as well as an illustrating example of the energy shift
in the degrader and an energy comparison at the surface.
Figure 4.1: The plots shows the lateral spread along the y axis of the different beam setups at
the entrance of the water phantom. By fitting the curves it to the gaussian curve, we achieve the
following values of the σy spread: Synchrotron: 0.58cm, ESS: 0.58cm, Degrader σE = 0.42 MeV:
4.36cm, Degrader σE = 0.42 MeV with Collimator: 0.41cm, Degrader σE = 1.4 MeV: 4.36cm
Figure 4.2: The plots shows the degradation of beam of the degrader setup through 27 cm PMMA.
The initial beam, the red line, has an energy of 250 MeV with an energy spread of 0.4 MeV. After the
degradation the beam, the green line, is shifted to a energy of 91.66 MeV and σE value of 3.09 MeV.
Figure 4.3: The plot shows the different energy spectra of the beams at the surface of the water
phantom, for a common range around 6.4 cm for all setups.
Table 4.1: The table shows the best fit approximation of the different spectra to the gaussian
distribution. All of the fits had an R2 value above .995.
Degrader σE = 0.42 MeV
Degrader σE = 0.42 MeV with Collimator
Degrader σE = 1.4 MeV
Figure 4.4: The figure shows the Bragg curves of the different initial setups, all with a range around
Table 4.2: The table shows the range (distal 90%) and the distal dose falloff (80% - 20%) of the
short range curves in figure 4.4.
Degrader σE = 0.42 MeV
Degrader σE = 0.42 MeV with Collimator
Degrader σE = 1.4 MeV
mean min max
Distal Falloff [mm]
mean min max
Figure 4.5: The figure shows the Bragg curves of the different initial setups, all with a range around
Table 4.3: The table shows the range (distal 90%) and the distal dose falloff (80% - 20%) of the long
range curves in figure 4.5.
mean min max
Degrader σE = 0.42 MeV
Degrader σE = 0.42 MeV with Collimator
Degrader σE = 1.4 MeV
Distal Falloff [mm]
mean min max
The distal dose falloff
Figure 4.6: The plot gives the full overview over the physical distal dose falloff for all clinical relevant
ranges. The curve fitting was done in to different ways. For the degrader setups, the horizontal line is
mean of all centre points: Degrader σE = 0.42 MeV: ddf (x) = 5.15mm, Degrader σE = 0.42 MeV with
Collimator: ddf (x) = 5.25mm, Degrader σE = 1.4 MeV: ddf (x) = 5.88mm. For the ESS/Synchrotron
points, the curve fitting was done by the best fit of a second order polynomial, passing through origo:
Synchrotron: ddf (x) = −3.6 × 10−7 x2 mm−1 + 0.0150x, ESS: ddf (x) = 5.8 × 10−7 x2 mm−1 + 0.0198x
Table 4.4: The table is based upon the approximation that the distal falloff for the degrader setups
is constant, independent of range. The mean and the standard deviation of the center points of the
different degrader setups, as well the standard deviation (std) and the margins with with the 95%
margins (mean ±2 std).
Degrader σE = 0.42 MeV
Degrader σE = 0.42 MeV with Collimator
Degrader σE = 1.4 MeV
Analysis and validity of model
ESS and synchrotron Beam
Even though the beam model might seem at bit to simple to give a realistic results, it is not completely
different from modern nozzle. The newest nozzles are made only with a scanning opportunity, which
means number of beam components are few as reasonable. Optimal the beam is only passing through a
couple of monitoring chambers, the glass of the vacuumized beam line and a short column of air, before
it hits the skin of the patient. Placing the source in air 1.5 metre in front of the patient and might
enlarge the beam spread beyond clinical realities.
Both of the fitted curves in figure 4.6 is close to proportional. This seems realistic, as the curves in
figure 2.1, also have an greater DDF for greater ranges.
The degrader setups in on the other hand a bit more complex, but not the great difference. PMMA was
chosen as the material for the degrader. This might not be possible to optimize, while other material,
as lead or other, is to dense and scatter the beam to much. Water could also be a possible degrader
material, but the nozzle layout rises in complexity if fluid water is added.
Two different set of beam parameters were chosen, both based upon published information about the
Mevion S250 cyclotron. Common for both system, is that the DDFBP is constant, and not depending
on the range of the beam. In other words, the beam inherit the DDFBP from the pure cyclotron
beam. The low uncertainty margins in table 4.4 also confirms this approximation. The degrader setup
with σE = 1.4MeV was based upon the first paper which published measurements of the depth dose
distribution of a momentum analyser free system. The σE was chosen as a best fit to the measurements,
so it could be thought that this is the right value of the energy spread from the cyclotron. In this paper
the DDFBP was stated to be ”nearly constant 6 mm“ for all configurations. In table 4.4 the simulated
DDFBP was found to be around 5.88 mm, which is near to 6 mm. 
Then it gives sense why the DDFSOBP is commented to be approximately 7mm, which gives a
ρ = 1.167, a reasonable ρ value. 
Effect of collimator
As seen in figure 4.1, the lateral spread of the degrader setups were far greater compared to the other
setups. A collimator was added to one of those, to rule out the effect of a larger focus. As seen in table
4.4, figure 4.6 and the other example figures, there is no significant difference between the system with
collimator compared to the system without. We could therefore preliminary rule out the correlation
between the beam size and the relative depth dose falloff. But other beam components might alter this,
so it is not a final conclusion.
The MAS is also able to filter the lateral spread of the beam. A nice lateral beam focus is obligatory for
an active scanning system, in case true IMPT is wanted. This issue, which also will have an dosimetric
impact, have not been investigated here, since it could possible be solved by technology. E. Pedroni
mentioned in a conference talk that the degrader should be placed as close to the patient, to minimize
the air gap and scattering. This fits with the drawing of the possible active nozzle layout in appendix
A.1.2, where the degrader is at the end of the nozzle, opposite to the passive scattering nozzle. He also
mentioned the possibility of adding an MLC as well, to collimate the outer regions of the field. 
Anyhow, it would be interesting to see how a the scattering problem will be handled by Mevion by
their new Hyperscan active scanning nozzle.
Other possible methods
The detector size might be far smaller. The optimal size of the diameter would be around 10 mm, similar
the sizes of clinical ionization chambers used for measuring the depth dose curve. The logging could be
done at lower resolution, since the position of d90, d80 and d20 is anyway linear interpolated. The offset
would be minimal up to around a resolution of at least 0.5 mm.
In this project, Monte Carlo simulations was chosen as the tool to measure the distal dose falloff
DDFBP . A better method would be to measure the falloff in a real facility, with a 250 MeV cyclotron
and a full ESS. A simple setup with a water phantom and a ionization chamber could be used to measure
the Bragg curves of a set of different energies from 70 MeV to 250 MeV. Analyse them and find their
DDFBP . Then a degrader system, a system without momentum analyser, could artificially be made
within the treatment room. A variable degrader could be made by plates of PMMA (or something
similar), and placed between the nozzle and the water phantom. The ESS should then set the energy
to 250 MeV and a constant σE , and the range could be by changed by changing the number of plates
in the beam. The Bragg curves could then be measured for many different ranges. While a system is
quite expensive, it would be cheaper just to use a TPS, where these measurements are already done for
existing systems. Models as the pencil beam model would then be sufficient to find a generic value of
DDFBP . Dropping the beam in a water phantom is the simplest of all experiments, and the answer
should not differ significantly from the Monte Carlo simulations.
A curve fitting could also be done, by fitting the distal curve to a normal distributed gaussian curve.
Then it also might be possible to find a analytical approximation of the distal dose falloff as a function
of E0 and σE , based upon the work done by Bortfeld. 
Dosimetric consequences at different depths
The graph are simple to read and understand. For the systems without a momentum analyser the
DDFBP are constant for all depths, while the DDFBP is proportional for the systems able to decrease
the σE .
The ESS line is cross the pure degrader system at R ≈ 25cm, but this does not directly implies that
the ESS system is worse for deeper tumours. It is fully dependent of the selection of the momentum slit
In general, we could conclude that the DDFBP are the same for the most deep seated tumours. By
comparing the degrader setup with σE = 0.42 to the synchrotron system, and setting a difference in
DDFBP below 2 mm as clinical neglectable, we could assume that both systems would able to give the
same treatment to tumours deeper than 21 cm within the body. According to table 2.1, both systems
would be able to treat prostate and rectum with the same degree of sparing dose to nearby OAR. the
two systems should also give an equal effect to some deep suited tumours within body, such as those in
the gastrological system.
On the other side, the DDFBP of the two different systems differ 3-5 mm for the most shallow
tumours. Eye tumours are some of the most shallow tumours treated with proton beam. With a
DDFSOBP of 7 mm, the brain and specially the visual nerve would most certainly get an unacceptable
amount of dose, even though this has not been directly calculated. The questions is then for which
other kinds of cancer that would not be suitable for treatment by a momentum analyser free system.
That questions is far above the goal of this thesis, but I would point out that it could be problematic to
treat some other shallow cancers, like skull chordom/chondrosarcom and breast, and maybe some cases
of head and neck and paediatric. To exemplify the possible limits of a system without MAS, a further
discussion of breast cancer is covered in subsection 5.2.4.
These problematic cases need to be considered and examined by a potential costumer of such a
Table 2.1 is in no way complete, and should be upgraded to cover the depth of all the Norwegian
patients group. This could be done by making proton plans of existing patients, and measure the distal
edge of the PTV of the treatment field chosen. This might needs to be done by some with experience in
proton therapy planning, whom is able to make realistic plans.
Constancy in treatment planning
In the conference proceeding paper, C. Bloch et al. discusses that the constancy of the Bragg peak shape
is a great advantage when it comes to treatment planning. In the TPS a single beam model would be
sufficient, instead of 24 distinct models for normal proton therapy centres. This is true fact, but dealing
with different beam models should not be a problem for a modern TPS. The difference in calculations
times should be minimal and neglectable, and the quality assurance of the beams could be automated.
I would therefore personally say that this statement is a false argument; a constant shape is not an
advantage in any way.
C. Bloch et al. also argues that even though there is a difference between the distal falloffs, this can not
be fully utilized, because of range uncertainties. A sub critical dose to the tumour is even worse than
extra dose to OAR, since the tumour might survive and the cells could be more radioimmune. Proton
therapy systems with a beam analyser will have need an extra margin beyond the distal edge of the CTV
to avoid that the tumour is given less than the prescribed dose. At MGH they typically add an extra
margin of 3.5% + 1 mm to compensate for the range uncertainties. Then every possible error is accounted
for, from miscalculations in the dose models to patient positioning. The uncertainty is dependent of the
total range of the SOBP, which means that the extra margin is minimal when shallower tumours are
treated. In figure 5.1 the typical margin added is listed for MGH and four different treatment centres in
Figure 5.1: The figure shows typical range uncertainty margin of five different treatment centres,
used in clinical routines today for the standard cases. The dashed line show the estimated uncertainty
of standard treatment, including all possible errors. The Solid line show an estimated for the more
complex geometries. The last dashed dotted line show the possible range uncertainty margin added,
if Monte Carlo simulations was utilized in the treatment planning. 
The range uncertainty could be minimized to 2.4% + 1.2 mm by including comprehensive Monte
Carlo calculations into clinical treatment planning. There are work to further minimalizing this margin,
by the help of promising image modalities, which may aid in the making better calculations, positioning
the patient and dose monitoring. Protongraphy and proton CT utilize the same principles as x-ray
radiography and CT, but delivers less dose to the patients. This could also give the true values for
proton absorption, some indications prove up to 5 % better values for the stopping power calculated
from Hounsfield units. Ongoing research also try to adapt offline PET scanners, to monitor the
fractured particles made by the proton beam. Some of these particles would be β-emitting isotopes, as
carbon-11, and could localize the impact of the proton beam. Other methods of monitoring the dose
is also investigated, including MRI, ultrasound and Cerenkov light.  . Recent developments
in prompt gamma-ray spectroscopy have made it possible to monitor the dose within 1.5 mm .
Figure 5.2: A comparison of two proton therapy dose plans of a lung tumour, as given in a a paper
by J. Flanz and T. Bortfeld. The plan to the left is an state-of-the-art plan in 2012 with a big ITV
and PTV to include for all movements and range uncertainties. The one to the right is the ultimate
goal of proton therapy, with a close to a perfect conformation of the PTV to the CTV.
The goal is a better conformation of the dose, as shown in a hypothetical dose plan in figure 5.2. J.
Flanz and T. Bortfeld named the possible evolution ”the third generation of proton therapy“. A quote
from their paper reads: ”Although reductions of uncertainty by factors of 2-6 are a formidable technical
challenge, there are good reasons to expect that substantial improvements are likely to be made in the
coming years. This should result in dose distributions no longer dominated by technological limitations
and determined purely by the laws of physics.“ ”Determined purely by the laws of physics“ means in our
case the physical limit of the DDFBP . And, as shown, the DDFBP is far superior for systems with a
momentum analyser compared to systems without, specially for shallow tumours. A system without a
momentum analyser therefore is not a system for the future, if high radiosurgerical precision is the goal.
Left Sided Breast Cancer: A clinical example
According to the Norwegian report of 2013, 2.1, between 140 and 180 breast cancer patients would
benefit from being treated with protons than x-rays. One of the biggest issues today in conventional
radiation therapy of breast cancer is the extra dose given to the lungs and the heart. This unwanted dose
is extra high for treating tumours in the left breast, at the same side as the heart. To minimize the extra
dose many clinics have implemented a treatment protocol as respiratory gating or tactical fixation of
the body and the breast, which both physically increase the distance between the treatment volume and
the cardiac volume. The dose to the left lung and the heart is further minimized by using two opposite
fields, tangential to the lung. This is shown by the yellow arrows in figure 5.3. 6 MV is usually the
photon energy of choice, to get the right build up under the skin. By using two opposite 6 MV fields the
dose is relative uniform for the longitudinal axis, so its possible to cover the whole planning treatment
volume by the prescribed dose within the acceptable limits.. Such a treatment plan utilize the steep
lateral penumbra of the photon beam to spare the lungs and the heart. High energy photons scatters
quite little, so the lateral penumbra is relative constant at different depths. According to Schlegel et al.
a standard linac with a integrated Multi Leaf Collimator (MLC) should have have a lateral penumbra
(80% - 20%) about 8-10 mm, when treating with 6 MV photons in soft tissue. But if the linac is
optimized with a µ-MLC, it is possible to get the lateral penumbra down to approximately 3 mm. In
the case of breast cancer, a lateral penumbra of 3 mm means that the dose to the left lung and heart is
already quite low. So if there should be any need for proton cancer, then this low unwanted dose should
be further lowered.
Figure 5.3: A axial plane drawing of the thorax. The red line shows the contour of the PTV of the
left sided breast cancer. The green lines outline the cardiac volume and the dark blue line the lungs.
Both of these are OAR in this treatment of breast cancer. The yellow lines shows a tangential fields,
used in external photon therapy to spare the OAR. The blue show a possible single field proton plan.
In treatment of breast cancer with protons, the most straightforward way would be to treat with
one field. The field should take the shortest way to the tumour, directly pointing at the breast, as
shown with the light blue arrow in figure 5.3. By using a SOBP to cover the full PTV, the left lung
and the heart would be on the backside of the treatment volume, in the beams eye view. In this case,
a steep falloff would be appreciated, specially if the whole breast tissue is regarded as tumour volume.
To simplify the problem, the distal edge of the PTV would be the same line as the inner line of the
tangential two field photon plan. By further simplification, we could then compare the dose to the lungs
and cardiac volume by comparing the lateral penumbra of the photon plan and the distal falloff of the
proton plan. Both parameters would then tell the distance which would be needed from going from a
high dose (80% of the prescribed) to a low dose (20% of the prescribed), if the dose was deposited in
a homogeneous medium. Depending on the size of the breast and the SOBP algorithm (ρ value), the
distal falloff from ESS/Synchrotron beam would be approximately 1-2 mm, the distal falloff of “pure
degrader system” would be 6-7 mm, and the lateral penumbra of the linac with µ-MLC would be 3 mm.
This is of course an idealized situation and it does not give the full picture of the cardiac dose. Anyway,
by comparing these parameters the “pure degrader system” is not just inferior to the ESS/Synchrotron
system, but also the optimized x-ray linac with µ-MLC. There is also the far less scattering and longer
particle penetration distance in the lungs, due to the low density and attenuation. This also speaks in
favour of the tangential photon plan, if the dose to the heart is considered. But the photon plan is only
able to draw a straight line behind the tumour, where the proton plan is able to conform the distal part
of the SOBP to the shape of the PTV, which would help to lower the cardiac dose.
But, as earlier mentioned, these arguments relies upon great simplifications. It is therefore strongly
recommended that the problem of left sided breast cancer should be look closer into by medical physicists
or dosimetrists, before any conclusions is made and systems bought. Optimized dose plans should be
made for all four setups, and there should be made a full comparison with dose. Due the inhomogeneous
region with lungs with low density, ribs with high density and tangential beams, the dose should be
calculated used a comprehensive dose calculation algorithm like the collapsed cone algorithm, or by
Monte Carlo simulations if possible. In current literature, no such clinical evaluation of breast cancer
plan have been found, only comparison with ESS systems.   It is not known to the author if
any breast cancer patients have been treated by the first and yet only Mevion therapy centre in St.
Louis, Missouri. According to preliminary data they have so far treated 70 patients, where 20% was
paediatric cases, 21% was lung and 6% was prostate. The news article says nothing about the 53% of
the patients. Mevion is also building up a new centre in Cleveland, Ohio. According to a business plan
of fall 2013, they where focusing on paediatric cases (35%), lung patients (15%), Brain/spine patients
(15%), cases of targeted re-irradiation or boost (15%), head and neck patients (10%) and GU/prostate
(10%). Breast cancer patients where not mentioned in the patient mix at all.
Market adaptations for manufactures
Even though the dose distribution might be slightly worse for a compact proton therapy system without
MAS compared to others, such a system might be well able to compete in a tough and small markets.
A full compact therapy system, will the full beam line and extra shielding, might be so expensive that
the economical cost-effect will be lower for such a system. So by cutting the cost to a minimum, such
a system might be favourable. In opposite to lower the lost, it could also focus on the deep tumours,
where the difference in the dose distribution is minimal. Prostate, as discussed in 5.2.1, is one of the
most common types of cancer, and the prevalence is rising as the general population gets older, so there
is a great volume. Prostate patients have also an income above average, so in private health markets the
demand would be higher. So the demand times the volume of prostate patients might be a good basis
for a business plan. Further it is also possible to get rid of the gantry, and only treat the patients with
a fixed horizontal beam. Prostate patients could be treat quite well with only two opposite field. The
patients would then be treated in a supine position, moved to the right position by a robotic table. In
such a setup, with a small cyclotron without any beam line and gantry, were almost reaching a room
size about the size of a linac bunker. Then the installation would also be far cheaper. By building
a smaller cyclotron, say about 160 MeV instead of 230-250 MeV, it is still possible to treat about 50 %
of the patient volume as shown in subsection 2.3.4. The distal dose falloff would then still be a little
higher for the shallowest tumours, but only at maximum 1-2mm. This might be within the acceptable
limits. There might also be possible to adjust the extraction energy of the cyclotron, instead of in
the degrader. Even though its is not yet done by any cyclotrons, but could work in theory. Mevion has
a patent of such a possible system. 
The reduction of the cost and size of compact proton therapy systems might also reduce the clinical
benefit of proton therapy for certain types of cancer. This work demonstrated the hypothesis that the
momentum analysis system has impact upon the distal dose falloff of the Bragg Curve, which hence
could also impact an extra dose to OAR close to the target volume. By dropping this, the distal falloff of
the deepest possible treatment range is inherited for all treatment depths. Compared to other systems
with a momentum analysis system, this inheritance might have dosimetric consequences which cannot
be neglected. This applies specially for shallow tumours, like eye, skull chordom/chondrosarcom and
But in this work, the DDFBP has been measured by using a simple model. While a greater DDFBP
in the simplified theory implies a worse dose distribution, this might not be true for a more complicated
clinical situation. I strongly recommended that this possible issue is further look in to, by comparing
dose plans of different systems, including dose plans of state-of-the-art photon therapy. The dose plans
should be made as realistic as possible to make a true comparison.
Commercial compact proton therapy
This appendix gives an overview over the different commercial companies which deliver a full standardized
compact single room centres to clinical costumers. All companies are active at the moment, and have
a compact single room solution among their options. They are all able to treat tumours at all depths
(proton energy at least 230 MeV) and from all angles (at least 180 degree rotatable gantry). Tailor
made solutions used by research centres are excluded from this list. The list might be incomplete by
some other minor suppliers at the market, like some Japanese companies. Table A.1 at the end gives an
summarized overview of these different compact systems.
Figure A.1: Overview over the Mevion S250 single room system. The cyclotron is vertically mounted
into the middle of massive 180 degree rotatable gantry. The superconducting synchrocyclotron directs
the beam directly into the nozzle. The system does therefore not not use a standard ESS, and degrades
the beam without filtering out neutrons and adjusting the momentum spread with a MAS. Currently
Mevion only have a passive scattering nozzle, but a active scanning nozzle, “Hyperscan“, would be
available in a couple of years. Courtesy of Mevion.
Figure A.2: A schematic diagram of the interior of the Mevion S250 passive scattering nozzle, as
drawn by P.M. Hill et al.. This is not to scale, but it gives a simple introduction of the different
components inside. The beam is leaving the accelerator and entering the nozzle to the left along
the axis. The beam is the scatter once, before it runs through the rotating modulator wheel. In
the Mevion passive scattering nozzle, the range is unconventionally shifted after the first scatter and
the modulator wheel. The second scatter flattens out the beam for the whole area, while the brass
aperture collimates the beam to shape of the tumour. This diagram also includes a water phantom.
All the coloured components in the diagram have to be changed for every field of every patient. It
should be noted that there is no distal compensator in this diagram, another personalized component
normally used in passive scattering technique. Courtesy of P.M Hill et al.. 
Figure A.3: A patent drawing over a possible construction of an active scanning nozzle, drawn by
Kenneth P. Gall et al.. Kenneth P. Gall was the founder and former CTO of Mevion Medical Systems.
The diagram shows rim of the Mevion cyclotron to the left (1005), the beam extraction channel (1002),
scanning magnets (1008) and an ion monitoring chamber (1009). The degrader (1010) is placed at the
end of the nozzle, straight in front of the patient, which would be situated to the right of the nozzle.
This degrader is consisting of controllable plates of material, able to move in and out of the beam.
Object 1028 is not mentioned in the patent, but the author would guess that could be some kind of a
collimator to constraint the lateral spread from the degrader. It is not yet known if the new Mevion
Hyperscan nozzle is based upon this configuration or not. Courtesy of K.P Gall et al.. 
Figure A.4: Overview over the new IBA Proteus ONE single room system. The accelerator is at
the back, the treatment room is in the middle and the control room is in the front. The is only able
to turn 220 degrees, so the room is open in the other part. IBA have for many years been the market
leader in the field of proton therapy. September 9. 2014 the first patient was treated by the first
compact Proteus ONE centre in Louisiana, Usa. Courtesy of IBA.
Figure A.5: Patent drawing of the new Proteus ONE with accelerator, beam line and gantry. The
most important parts is the accelerator (40), the degrader (41), and the gantry (15). The ESS is
integrated as a part of the gantry with the focusing quadrupole magnets (44), the beam collimators
(42 and 45), the momentum slit (43) and the bending magnets (47-49). Since a lot of neutrons is
produce in the degrader, an extra neutron absorber (51) is added upstream to protect the patient.
The new superconducting synchrocyclotron IBA S2C2 is special made for the new Proteus ONE, to
save space and energy compared to the regular C235 isocyclotron used in multi room systems. The
new IBA gantry have an ESS built in as a part of the gantry, which are fully exploiting the beam
bending magnets which are anyway there. This is an other elegant method to save space, since the
accelerator could be place right behind he gantry. Courtesy of IBA 
Probeam Compact Proton Therapy System
Figure A.6: A patent drawing of the new Varian ProBeam compact Proton Therapy system. The
accelerator is to the left, the beam line with the degrader is in the middle and the gantry and treatment
room is to the right. Varian follows the same principle of integrating the ESS into the gantry, by
utilizing the dipole magnets which are already there. Varian already have a compact superconducting
isocyclotron in their multi room centres around the world, which they also plan to use for the compact
centres. In opposite to Mevion, IBA and Protom, Varian still wants to stick to the full 360 degrees
gantry for the single room system. There is still not built a single room system like this yet. Courtesy
of Varian 
Cyclotron and ESS mounted on gantry
Figure A.7: A computerized drawing of a conceptual compact single room system, made by Accel
(today known as Varian). It shows a fully rotatable gantry with the “treatment room” in the front
with the nozzle and treatment table. The ProBeam superconducting isocyclotron is placed in the back
of the gantry, as a counterweight. At the top, between the cyclotron and the nozzle, the beam line
is bent 180 degrees and fully equipped as a normal ESS with dipoles, quadropoles and collimators.
There is no other beam line outside the gantry, everything is rotating together. This is a possible way
to include a ESS in a gantry mounted cyclotron system. But Varian newer constructed this system,
most certainly because such a system is far to heavy and expensive. Courtesy of Varian 
Figure A.8: A drawing of a two story single room system, made by Sumitomo Heavy Industries.
The cyclotron is placed in the basement of the building and the gantry on the ground floor, with a
vertical beam line and ESS between. By using a “double-decker” system, the spatial footprint is still
quite low. In contrast to IBA, Varian and Mevion, Sumitomo does not use a small superconducting
cyclotron in their compact therapy system. It would not save any space by using a superconducting
cyclotron, if they stick to building a two story system. The gantry is a special “corkscrew“ gantry,
which is shorter than conventional gantries. Sumitomo have built the world first two story centre in
Nagano. At the moment they are also building the first expandable single room system in Sapporo.
Courtesy of Sumitomo Heavy Industries.  
Figure A.9: A drawing of a two room system, made by Pronova. The cyclotron is placed in the
same floor as the treatment rooms. The gantries are made smaller than other suppliers, due to the use
of strong superconducting magnets. Every treatment room is equipped with its own ESS. Pronova is
installing their first centre these days, a two room system in Knoxville, Tennessee, USA. They also
offer a single room system, which is simple to be expand, due to the designated ESS for every gantry.
Courtesy of Pronova 
Figure A.10: The computerized drawing shows a possible setup of a Protom Radiance 330 compact
single room system. In this setup, the accelerator is placed in the story over the treatment room,
which gives a the possibility for a small footprint. The Protom Radiance 330 is based upon a small
synchrotron, available to accelerate protons all the way up to 330 MeV, making it suitable for protongraphy and proton computerized tomography of all parts of body. By using a synchrotron, it is
possible to extract the beam at any energies, which excludes the need for an ESS. Protom also delivers
a 180 degree scanning gantry to their locations. At the moment, they are building multi room centre
in Flint, Michigan in US, and have sold a single room centre to Massachusetts General Hospital in
Boston, US. Drawing courtesy of Protom International.
Figure A.11: The drawing shows a possible setup of a Mitsubishi compact single room system. Here
the accelerator is placed in the same story as the treatment room. Mitsubishi have built 10 synchrotron
based particle centres, including HIMAC, the first clinical centre treating with carbon ions. Anyhow,
they still have no centres outside of Japan. Mitsubishi have specialized in synchrotrons, and therefore
have no need for an ESS. Mitsubishi normally delivers full 360 degrees gantries. Drawing courtesy of
Mitsubishi Particle Therapy System.
1 R, 1 P
Commercial model name
Compact systems in operation
Room layout (Rooms, Planes)
Beam delivery method
Expandable to multiple rooms
Momentum analysis system
2 R, 1 P
2 R, 2 P
2 R, 1 P
2 R , 1/2 P
Table A.1: The table gives an overview over the different compact proton therapy systems that are
currently available on the open market. Based on an old table from Aarhus report. 
2 R, 1/2 P
 W.C. Röntgen, On a new kind or rays. Journal of the Franklin Institute 141.3 183-191 (1896)
 Robert L. Bowles, Pathological and therapeutic value of the roentgen rays. The Lancet 147 655-656
 E.H. Grubbé, Priority in the therapeutic use of X-rays. Radiology 21 156-162 (1933)
 A.H. Becquerel, Sur les radiations émises par phosphorescence. Compt. Rend. Acad. Sci. 122, 420421 (1896)
 J.M. Slater, From X-Rays to Ion Beams: A Short History of Radiation Therapy Ion Beam Therapy:
Fundamentals, Technology, Clinical Applications, Springer 3-16 (2012)
 W.H. Bragg and R. Kleeman, LXXIV. On the ionization curves of radium. The London, Edinburgh,
and Dublin Philosophical Magazine and Journal of Science 8.48, 726-738 (1904)
 R.R. Wilson, Radiological use of fast protons. Radiology 47.5 487-491 (1946)
 W.P. Levin, H. Kooy, J.S. Loeffler and T.F. DeLaney Proton beam therapy. British Journal of Cancer
93 849–854 (2005)
 M. Jermann Hadron Therapy Patient Statistics of March 2013 http://www.ptcog.ch (2013)
 Planlegging av norsk senter for partikkelterapi http://www.medfys.no (2013)
 J.M. Schippers and A.J. Lomax, Emerging technologies in proton therapy. Acta Oncologica 50.6
 C. Bloch et al., Startup of the Kling Center for Proton Therapy in AIP Conference Proceedings
1525 314-318 (2013)
 P.M Hill et al., Optimizing field patching in passively scattered proton therapy with the use of beam
current modulation Phys Med Biol. 58.16 5527-5539 (2013)
 W.H. Bragg and R. Kleeman, XXXIX. On the α particles of radium, and their loss of range in
passing through various atoms and molecules. The London, Edinburgh, and Dublin Philosophical
Magazine and Journal of Science 10.57 318-340 (1905)
 N. Bohr, II. On the theory of the decrease of velocity of moving electrified particles on passing through
matter. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science 25.145
 H. Bethe and W. Heitler, On the stopping of fast particles and on the creation of positive electrons.
Proceedings of the Royal Society of London. Series A 146.856 83-112 (1934)
 B. Gottschalk, Physics of Proton Interactions in Matter Proton Therapy Physics, CRC Press/Taylor
& Francis. 19-60 (2012)
 T. Bortfeld, An analytical approximation of the Bragg curve for therapeutic proton beams. Medical
physics 24.12 2024-2033 (1997)
 E. Pedroni, Experimental characterization and physical modelling of the dose distribution of scanned
proton pencil beams. Phys. Med. Biol. 50 541-561 (2008)
 U. Linz, Physical and Biological Rationale for Using Ions in Therapy Ion Beam Therapy: Fundamentals, Technology, Clinical Applications, Springer 45-60 (2012)
 L. Grzanka, Modelling beam transport and biological effectiveness to develop treatment planning for
ion beam radiotherapy Doctoral dissertion (2013)
 H. Paganetti, Relative biological effectiveness (RBE) values for proton beam therapy. Variations as
a function of biological endpoint, dose, and linear energy transfer. Phys. Med. Biol. 59 R419 (2014)
 H. Paganetti, Range uncertainties in proton therapy and the role of Monte Carlo simulations Phys.
Med. Biol. 57 R99 (2012)
 O. Jäkel, Medical physics aspects of particle therapy. Radiation protection dosimetry 137.1-2 156166 (2009)
 M.G. Holloway, and M.S. Livingston, Range and specific ionization of alpha-particles. Physical
Review 54.1 18 (1938)
 R. Wideröe, New Principle for Production of High Potentials. Archiv für Elektrotechnik 21 (1928)
 E.O. Lawrence and N.E. Edlefsen., On the production of high speed protons. Science 72.1867 376
 U. Amaldi, Cancer therapy with particle accelerators. Nuclear physics A 654.1 C375-C399 (1999)
 B. Gottschalk, Passive Beam Scattering Proton and Charged Particle Therapy Radiotherapy, Lippincott Williams & Wilkins 33-40 (2008)
 S.A. Kostromin, E.M. Syresin, Trends in accelerator technology for hadron therapy Phys. Part.
Nuclei Lett. 10 833–853 (2013)
 J.M. Slater, et al, The proton treatment center at Loma Linda University Medical Center: rationale for and description of its development. International Journal of Radiation Oncology* Biology*
Physics 22.2 383-389 (1992)
 H.M. Lu and J. Flanz, Characteristics of Clinical Proton Beams Proton Therapy Physics, CRC
Press/Taylor & Francis 103-124 (2012)
 T. Kanai, et. al, Spot scanning system for proton radiotherapy. Medical physics 7.4 365-369 (1980)
 E. Pedroni, Active Beam Scanning Proton and Charged Particle Therapy Radiotherapy, Lippincott
Williams & Wilkins 40-49 (2008)
 D.S. Parcerisa, Experimental and computational investigations on the water-to-air stopping power
ratio for ion chamber dosimetry in carbon ion radiotherapy Doctoral dissertation, University of
 M. Goitein, et al, Multi-dimensional treatment planning: II. Beam’s eye-view, back projection, and
projection through CT sections. International Journal of Radiation Oncology* Biology* Physics 9.6
 D. Verellen, M.E. Ridder, and G. Storme, A (short) history of image-guided radiotherapy. Radiotherapy and Oncology 86.1 4-13 (2008)
 W. Parker and H. Patrocinio, Clinical treatment planning in external photon beam radiotherapy.
Radiation Oncology Physics Handbook 219-272 (2005)
 H. Suit, et al. Should positive phase III clinical trial data be required before proton beam therapy is
more widely adopted? No. Radiotherapy and Oncology 86.2 148-153 (2008)
 M. Goitein and M. Jermann, The Relative Costs of Proton and X-ray Radiation Therapy Clin.
Oncol. 15 537-550 (2003)
 A. Konski, et. al, Is proton beam therapy cost effective in the treatment of adenocarcinoma of the
prostate?. Journal of Clinical Oncology 25.24 3603-3608 (2007)
 T. Björk-Eriksson and B. Glimelius, The potential of proton beam therapy in paediatric cancer. Acta
Oncologica 44.8 871-875 (2005)
 K.P. Gall, The Single-Room Ion Beam Facility Ion Beam Therapy: Fundamentals, Technology,
Clinical Applications, Springer 661-717 (2012)
 ICRU, Beam Delivery and Properties, Journal of the ICRU 7 29–48 (2007)
 K. Parodi, Monte Carlo Methods for Dose Calculations Ion Beam Therapy: Fundamentals, Technology, Clinical Applications, Springer 97-116 (2012)
 U. Oelfke and C. Scholz, Dose Calculation Algorithms New Technologies in Radiation Oncology,
Springer 187-196 (2006)
 T. Haberer et al., Ion beam therapy system and a method for operating the system. Google Patents,
 J.M. Schippers et al., The use of protons in cancer therapy at PSI and related instrumentation. J.
Phys.: Conf. Ser. 41.6 61 - (2006)
 J. Cameron and N. Schreuder, Smaller – Lighter – Cheaper: New Technological Concepts in Proton
Therapy Ion Beam Therapy: Fundamentals, Technology, Clinical Applications, Springer 673-686
 K. Parodi and W. Enghardt, Potential application of PET in quality assurance of proton therapy.
Phys. Med. Biol. 45.11 N151-6 (2000)
 K. L. Chen et al., Evaluation of neutron dose equivalent from the Mevion S250 proton accelerator:
measurements and calculations. Phys. Med. Biol. 58.24 8709 - 8723 (2013)
 J. Flanz and T. Bortfeld Evolution of Technology to Optimize the Delivery of Proton Therapy: The
Third Generation. Seminars in Radiation Oncology, Controversies in Proton Therapy 23 142–148
 J. Dineley, Proton CT improves stopping power accuracy http://medicalphysicsweb.org (2014)
 K. Parodi, et. al, PET/CT imaging for treatment verification after proton therapy: a study with
plastic phantoms and metallic implants. Medical physics 34.2 419-435 (2007)
 B.W. Raaymakers, A.J.E. Raaijmakers, and J.J.W. Lagendijk, Feasibility of MRI guided proton
therapy: magnetic field dose effects. Physics in medicine and biology 53.20 5615 (2008)
 Y. Helo et al, The physics of Cerenkov light production during proton therapy. Phys. Med. Biol. 59
 T. Freeman, Ionoacoustics: a new take on range verification http://medicalphysicsweb.org (2014)
 J.M Verburg and J. Seco, Proton range verification through prompt gamma-ray spectroscopy. Phys.
Med. Biol. 59 7089 (2014)
 B. Fowble et al., Cancers of the Breast Leibel and Phillips Textbook of Radiation Oncology, third
edition, Elsevier & Saunders 1215-1322 (2010)
 W. Schlegel et al., Beam Delivery in 3D Conformal Radiotherapy Using Multi-Leaf Collimators New
Technologies in Radiation Oncology, Springer 257-266 (2006)
 J. Johansson et al, Node-positive left-sided breast cancer patients after breast-conserving surgery:
potential outcomes of radiotherapy modalities and techniques. Radiotherapy and Oncology 65 89-98
 D.C. Weber et al, Radiation therapy planning with photons and protons for early and advanced breast
cancer: an overview Radiation Oncology 1 22 (2006)
 Newsarticle from Astro, September 2014 http://www.advisory.com/research/marketing-andplanning-leadership-council/the-growth-channel/2014/09/astro-2014-the-future-of-proton-therapy
 B. W. Wessels, Planning for the Installation of a Compact Superconducting Proton Therapy Unit
in a Busy Medical Center Environment Penn-Ohio Chapter of AAPM, 2013 Fall Symposium,
 E. Pedroni Accelerator and Beamline/Gantry Design from a user’s point of view - for optimal beam
delivery Conference talk at PTCOG 52 (2013)
 G. T. Zwart et al., Adjusting energy of a particle beam US Patent App. 14/039,342 (2014)
 Homepage of Mevion Medical Systems http://www.mevion.com (2014)
 Y. Jongen et al., Gantry comprising beam analyser for use in particle therapy WO Patent App.
 IBA Proteus ONE Brochure http://www.iba-protontherapy.com/ (2014)
 M. Benna et al., Compact proton therapy system with energy selection onboard a rotatable gantry
US Patent App. 14/033,950, https://www.google.com/patents/US20140275699 (2014)
 Homepage of Light Ions, Sumitomo Heavy Industries reseller in Europe and America
 T. Morita and T. Tachikawa Performance of Double-Decker compact proton therapy system ESTRO
33 Conference, EP-1782 (2014)
 Homepage of Pronova http://www.pronovasolutions.com (2014)
 Homepage of Protom International http://www.protominternational.com (2014)
 Homepage of Mitsubishi Particle Therapy System
 The Danish National Center for Particle Radiotherapy http://www.rm.dk/files/ (2011)