Design and commissioning of the non-dedicated scanning proton beamline

for ocular treatment at the synchrotron-based CNAO facility

Mario Ciocca, Giuseppe Magro, Edoardo Mastella,a) Andrea Mairani, Alfredo Mirandola,
Silvia Molinelli, Stefania Russo, Alessandro Vai, and Maria Rosaria Fiore
Fondazione CNAO, strada Campeggi 53, 27100 Pavia, Italy
Carlo Mosci
Ente Ospedaliero Ospedali Galliera, via Mura delle Cappuccine 14, 16128 Genova, Italy
Francesca Valvo
Fondazione CNAO, strada Campeggi 53, 27100 Pavia, Italy
Riccardo Via
Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano, piazza Leonardo da Vinci 32, 20133 Milano,
Italy

Guido Baroni
Fondazione CNAO, strada Campeggi 53, 27100 Pavia, Italy
Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano, piazza Leonardo da Vinci 32, 20133 Milano,
Italy
Roberto Orecchia
Fondazione CNAO, strada Campeggi 53, 27100 Pavia, Italy
Istituto Europeo di Oncologia, via Ripamonti 435, 20100 Milano, Italy
(Received 7 September 2018; revised 11 December 2018; accepted for publication 9 January 2019;
published xx xxxx xxxx)
Purpose: Only few centers worldwide treat intraocular tumors with proton therapy, all of them with
a dedicated beamline, except in one case in the USA. The Italian National Center for Oncological
Hadrontherapy (CNAO) is a synchrotron-based hadrontherapy facility equipped with fixed beamlines
and pencil beam scanning modality. Recently, a general-purpose horizontal proton beamline was
adapted to treat also ocular diseases. In this work, the conceptual design and main dosimetric proper-
ties of this new proton eyeline are presented.
Methods: A 28 mm thick water-equivalent range shifter (RS) was placed along the proton beamline
to shift the minimum beam penetration at shallower depths. FLUKA Monte Carlo (MC) simulations
were performed to optimize the position of the RS and patient-specific collimator, in order to achieve
sharp lateral dose gradients. Lateral dose profiles were then measured with radiochromic EBT3 films
to evaluate the dose uniformity and lateral penumbra width at several depths. Different beam scan-
ning patterns were tested. Discrete energy levels with 1 mm water-equivalent step within the whole
ocular energy range (62.7–89.8 MeV) were used, while fine adjustment of beam range was achieved
using thin polymethylmethacrylate additional sheets.
Depth-dose distributions (DDDs) were measured with the Peakfinder system. Monoenergetic beam
weights to achieve flat spread-out Bragg Peaks (SOBPs) were numerically determined. Absorbed
dose to water under reference conditions was measured with an Advanced Markus chamber, follow-
ing International Atomic Energy Agency (IAEA) Technical Report Series (TRS)-398 Code of Prac-
tice. Neutron dose at the contralateral eye was evaluated with passive bubble dosimeters.
Results: Monte Carlo simulations and experimental results confirmed that maximizing the air gap
between RS and aperture reduces the lateral dose penumbra width of the collimated beam and
increases the field transversal dose homogeneity. Therefore, RS and brass collimator were placed at
about 98 cm (upstream of the beam monitors) and 7 cm from the isocenter, respectively. The lateral
80%–20% penumbra at middle-SOBP ranged between 1.4 and 1.7 mm depending on field size,
while 90%–10% distal fall-off of the DDDs ranged between 1.0 and 1.5 mm, as a function of range.
Such values are comparable to those reported for most existing eye-dedicated facilities. Measured
SOBP doses were in very good agreement with MC simulations. Mean neutron dose at the contralat-
eral eye was 68 lSv/Gy. Beam delivery time, for 60 Gy relative biological effectiveness (RBE) pre-
scription dose in four fractions, was around 3 min per session.
Conclusions: Our adapted scanning proton beamline satisfied the requirements for intraocular tumor
treatment. The first ocular treatment was delivered in August 2016 and more than 100 patients suc-
cessfully completed their treatment in these 2 yr. © 2019 American Association of Physicists in Med-
icine [https://doi.org/10.1002/mp.13389]

1 Med. Phys. 0 (0), xxxx 0094-2405/xxxx/0(0)/1/xx © 2019 American Association of Physicists in Medicine 1
2 Ciocca et al.: Scanning proton ocular beamline
Key words: commissioning, ocular treatment, pencil beam scanning, proton therapy, uveal
melanoma
1. INTRODUCTION
Radiation therapy nowadays represents the standard conserv-
ing approach to treat uveal melanoma and other less common
ocular tumors, such as retinoblastomas and choroidal metas-
tases.1,2 Different types of brachytherapy (106-Ruthenium
and 125-Iodine plaques) and external radiotherapy (photon
stereotactic radiosurgery, proton beams) are used for high
tumor control and globe preservation, alternative to enucle-
ation. However, proton beam radiotherapy (PBT) is generally
preferred since it can provide more favorable results in terms
of vision retention and side effects.1,3–8 As assessed by two
recent surveys, respectively, performed at the European level
and worldwide,1,9 a limited number of centers (around fif-
teen) in Europe and North America can provide ocular PBT.
All of them use a cyclotron to produce proton beams at high-
dose rates, mostly with dedicated horizontal fixed beamlines,
passive scattering modality and single-anterior beam
approach.1,10 Only one experience on a non-dedicated nozzle
for ocular PBT, using image-based three-dimensional (3D)
treatment planning and multiple beams, has been recently
reported.11
In Italy, patients affected by ocular lesions and eligible for
PBT (a few hundreds per year) so far have been sent abroad
for treatment, mainly to France12 or Switzerland, or treated at
CATANA (Laboratori Nazionali del Sud, INFN, Catania,
Italy)13,14 to a lesser extent. In 2011, the National Center for
Oncological Hadrontherapy (CNAO), first hospital-based
synchrotron facility built in Italy, started to treat patients with
scanning high-energy proton and carbon ion beams.15–17 Four
general-purpose fixed beamlines (three horizontal and one
vertical) in three different treatment rooms are available to
treat deeply seated tumors mostly in the head and neck (such
as skull base), abdominal and pelvic regions. Following the
start-up phase when several hundreds of patients were treated,
CNAO was asked by the Italian Ministry of Health to investi-
gate the feasibility to extend clinical activity to ocular treat-
ments. However, the option of temporary stopping the
clinical routine in one of the treatment rooms and converting
the beamline to be fully dedicated to eye treatments was
judged as unsustainable in the context of the global strategy
of the center. Therefore, we started to investigate how to
adapt the existing, general-purpose horizontal beamlines,
designed for pencil beam scanning (PBS) modality, to also
meet the specific requirements of proton ocular treatments, in
the challenge of being able to rapidly switch machine settings
and treatment accessories between ocular and non-ocular
patients (these latter representing the large majority of cases).
A preliminary study based on Monte Carlo (MC) simula-
tions showed that the starting strategy of minimizing air gaps
among patient skin (ocular globe), range shifter (RS) and
nozzle, which is commonly valid for conventional proton
therapy,18,19 lead to very poor results, in terms of
Medical Physics, 0 (0), xxxx
2
unacceptable lateral beam penumbra width, even in the pres-
ence of a field-specific ocular collimator.20 That finding led
us to investigate completely different approaches, as reported
in this work.
The aim of this paper was to present the conceptual design
and results of dosimetric commissioning of the first world-
wide non-dedicated proton beamline based on PBS modality
and single-field irradiation for ocular treatment, developed at
CNAO.
Details of the ocular treatment process and specific clini-
cal workflow, including patient selection criteria, treatment
simulation, planning, verification and delivery, are out of the
scope of this paper and will be presented separately.
2. MATERIALS AND METHODS
2.A. General-purpose CNAO beamlines and
specificities of the proton eyeline
The design and main standard components of the CNAO
fixed beamlines, including synchrotron, scanning magnets
for PBS, dose delivery system (DDS), the patient positioning
system and the x-ray patient verification system (PVS) have
been already described elsewhere.15–17 The isocenter and
beam central axis are identified by means of room lasers, as
shown in Fig. 1. Additionally, a custom eye tracking system
(ETS) was specifically developed for gazing direction fixa-
tion and real-time eye motion detection.21
At CNAO, proton (carbon ion) beams can be extracted
from the synchrotron with discrete energies ranging between
62.7 MeV (115.2 MeV/u) and 228.6 MeV (398.8 MeV/u),
corresponding to a Bragg peak (BP) depth in water from 3 to
32 (27) cm, with a water-equivalent step of 2 mm. To treat
tumors located at depths shallower than 3 cm, RSs of differ-
ent thicknesses, consisting of water-equivalent RW3 slabs
(PTW, Freiburg, Germany), are currently used.
In the perspective of treating eye tumors, additional proton
energies have been commissioned, up to 89.8 MeV, to
increase the energy resolution to 1 mm water-equivalent step.
Moreover, for ocular treatments requiring very fine adjust-
ments of beam range, thin certified polymethylmethacrylate
(PMMA) additional dedicated sheets (thickness = 0.175 mm,
Goodfellow Cambridge Limited, Huntingdon, UK) were used.
The relative water-equivalent path lengths (rWEPLs) of
the two different types of RSs were determined from BP
curves, acquired using the PTW Peakfinder water column, as
described in Mirandola et al.16
A patient-specific brass collimator (or aperture) for eye
treatments was also adopted to reduce the beam lateral
penumbra (LP), thus sparing as much as possible the dose to
the organs at risk close to the target, for example, lens, mac-
ula and optic disk. Brass type with the minimum amount of
lead (type OT36) was chosen in order to minimize the
3 Ciocca et al.: Scanning proton ocular beamline
FIG. 1. The CNAO eyeline with its main elements: (a) vacuum window; (b) 27 mm thick RW3 range shifter; (c) dual beam monitor chambers; (d) additional
polymethylmethacrylate thin sheets; (e) RW3 beam pre-collimator; (f) individualized brass collimator; (g) isocenter, identified by laser red lines. Beam comes
from left. A thermoplastic mask for patient fixation, attached to the treatment chair, is also shown on the right. Distances of (b) and (f) from the isocenter are 98
and 7 cm, respectively.
production of secondary neutrons. The brass thickness
needed to achieve a beam attenuation higher than 99% was
again determined using the Peakfinder system, together with
MC simulation used to validate material composition settings
within the code itself.
Once separately characterized, all those passive compo-
nents were simultaneously tested on the beamline with a
combined MC and experimental approach, to find out their
optimal positioning along the beam path.
2.B. Monte Carlo optimization of the eyeline
geometry
The FLUKA MC code22–24 was used to support the start-
up and the clinical operation of the new CNAO proton eye-
line. The full CNAO horizontal beamline, including the vac-
uum window, beam monitoring (BM) system, and all passive
elements, was accurately modeled using the FLUKA combi-
natorial geometry according to the design details.24,25 By fol-
lowing the same strategy reported in Parodi et al.,26 a set of
low-energy proton beams was commissioned with nominal
BPs 1 mm stacking in depth.
MC simulations of the transmission coefficient were per-
formed to determine the optimal thickness of the patient-spe-
cific collimator, by using a 65 MeV proton beam (32 mm of
BP depth in water).
For a better understanding of the interactions between
scanning proton beams and both RS and collimator
mounted along the beamline, initial tests were performed
for an elementary proton pencil beam and simplified
beamline geometry.
The lateral dose distribution of a 90 MeV proton pencil
beam (60 mm of BP depth in water), with the energy reduced
by the insertion of a 30 mm thick water absorber, was scored
just prior to the 15 mm diameter collimator, as well as 5 cm
downstream of it. The RS-induced modification to the beam
Medical Physics, 0 (0), xxxx
3
width (prior and after collimation) was assessed as a function
of the air gap between those two passive elements. An initial
full width at half maximum (FWHM) of 4 mm was set, com-
parable to the one available at CNAO at the vacuum window.
Radial dose profiles were scored for a number of RS-collima-
tor air gaps, covering a wide range, from 2 to 82 cm.
A proper collimator-to-skin distance minimizing the pro-
ton scattering was also investigated, using the same simpli-
fied geometry.
Based on MC simulation (see Section 3.A), the final con-
figuration of the eyeline adopted at CNAO is shown in Fig. 1,
starting from the vacuum window (a). To maximize the air
gap, the RS (b) is placed as close as possible to (a), that is
upstream of the BM chambers (c), thus strongly influencing
their response (see Section 3.D). The RS is mounted on a
removable frame. A fixed thickness of the RS equal to
27 mm of RW3 (28.1 mm water-equivalent) was chosen,
hence achieving a penetration range between 2 and 32 mm in
water for the lower set of beam energies produced by the syn-
chrotron (62.7 to 89.8 MeV).
A removable applicator mounted onto the nozzle, just
downstream of the BM chambers, holds PMMA additional
sheets (d) used (when needed) to finely achieve the patient-
specific planned range, beam pre-collimator (e) used to
absorb peripheral dose halo (see Section 3.C) and finally the
individual collimator (f) close to the isocenter (g).
2.C. Depth-dose distributions and SOBP generation
For beam commissioning purposes, laterally integrated
depth–dose distributions (DDDs) were measured with the
PTW Peakfinder water column for a set of representative
mono-energetic unscanned uncollimated pencil beams, as
described in Mirandola et al.16 The acquired curves were
compared against MC simulations performed by scoring
deposited energy in a cylinder of 4.08 cm radius
4 Ciocca et al.: Scanning proton ocular beamline
(corresponding to the radius of the PTW BP ionization cham-
ber, model TM34080, mounted in the Peakfinder system).
Full MC database for all the available energies was then gen-
erated, under the eyeline configuration. Square mono-ener-
getic scanned fields were simulated, using 3 9 3 cm2 as
nominal field size (FS) at the isocenter and 2 mm as transver-
sal scanning step, for a total number of 225 equally weighted
spots. Dose was scored on a 2.5 9 2.5 9 5 cm3 water mini-
phantom with a dose-grid resolution of 0.1 mm, with its
entrance surface placed 15 mm upstream of the isocenter.
A set of 31 energy-specific 3D-matrices reporting absorbed
dose per unit primary weight was stored and postprocessed to
retrieve the required beam configuration data, that is, proximal,
distal and LP profiles. To this purpose, a MATLAB-based
software (version R2013a, The MathWorks Inc., Natick, MA,
USA) was developed, which translates treatment plan parame-
ters (i.e., range and extent of modulation) into machine set-
tings, that is, beam energy, beam relative weights and proper
amount of thin absorber sheets (in addition to the main thick
RS already mentioned) to guarantee flat SOBPs. The calcula-
tion-engine of that code is based on a least-squares method as
proposed by Lomax.27 An off-line mask, acting on the 3D-
matrices, restricts the database handled by the iterative algo-
rithm to a cylindrical scoring mesh matching the sensitive area
(radius = 2.5 mm) of the reference ionization chamber fore-
seen for use in the clinical practice (PTW Advanced Markus
chamber, type 34045), in order to characterize the resulting
DDDs, which are then used for clinical SOBP generation.
SOBPs with different ranges and modulations were then
calculated. Following beam monitor chamber calibration
(see Section 2.E), absorbed doses were measured in a water
phantom (PTW type 41023) at different depths along the
SOBPs with the reference Advanced Markus chamber and
compared with MC simulations in relative terms. The use of
the PTW proton diode (type PR60020) or a micro-diamond
detector28–32 is foreseen for FSs below 1 cm (i.e., less than
twice the sensitive area of the ionization chamber, as recom-
mended in International Atomic Energy Agency (IAEA)
Technical Report Series (TRS)-39833).
2.D. Transversal dose profiles
Field dose uniformity and LP width for the eyeline setup
were evaluated using radiochromic EBT3 films (Ashland Inc.,
Bridgewater, NJ, USA), placed in an RW3 slab phantom. Once
irradiated, films were scanned in transmission mode using a
commercial flat-bed scanner, at a spatial resolution of about
0.2 mm/pixel (150 dpi). Red-green-blue images were collected
at 48 bits (16 bits per color channel). Red channel signal
extraction from scanned image and film analysis were per-
formed using the PTW MEPHYSTO mc2 commercial soft-
ware. Single calibration curve, independent of proton energy
and obtained by irradiating a set of film strips from the same
lot perpendicularly to the beam at dose levels of 0–10 Gy, was
used to convert optical density into dose.10,16
Lateral dose profiles were analyzed to evaluate FS, LP,
field flatness and symmetry, as defined in ICRU Report
Medical Physics, 0 (0), xxxx
4
no. 7834. Flatness and symmetry were calculated in the 80%
central region of the profile for FSs larger than 15 mm. For
smaller fields, to keep the exclusion of LP from uniformity
analysis, those two parameters were calculated within the so-
called target width,34 here defined as the FS minus 4 mm
(these latter approximately representing twice the 90%–50%
LP width for each side of the profile).
Firstly, films were irradiated orthogonally to the beam
direction at the depth of 15 mm with a square mono-energetic
(81.6 MeV) uniformly scanned field, at doses around 4–
6 Gy. The reference 15 mm diameter aperture was used. Dif-
ferent scanning FSs and spot spacing were tested. Proper lat-
eral extent of the brass block to avoid unwanted peripheral
dose halo to the patient was also determined by exposing a
large-area film (20 cm 9 25 cm) to uncollimated scanning
beam.
Then, the LP dependence on depth in water was investi-
gated with the MC code for several SOBPs, different in
terms of range and modulation, and the 15 mm diameter
collimator. As a matter of comparison, experimental mea-
surements with EBT3 films were also performed at the
depths of 7 and 17 mm, for an SOBP of 20 mm range and
15 mm modulation.
Finally, uniformity and LP were measured with films for
seven collimators having different shape (circular or ellipti-
cal) and FS, from 6 to 25 mm. In particular, four collimators
represented real cases of patients treated at our center. Again,
films were placed inside an RW3 plastic phantom at the depth
of 15 mm and were irradiated with a typical SOBP (25 mm
range and 21 mm modulation).
2.E. Calibration of the beam monitor chambers
Once the optimal setup of the eyeline was defined, the
absorbed dose to water under reference conditions was mea-
sured and calibration of BM chambers performed as follows.
Dose determination was based on the IAEA TRS-398 Code
of Practice for proton beams, using a PTW Advanced Markus
ionization chamber (type 34045) and 15 mm diameter aper-
ture as a reference FS.33 BM calibration curve dedicated to
ocular treatment was required, since the optimal position of
the main RS was found to be upstream of the BM chambers
(with respect to beam direction), rather than downstream as
normally happens.16
The calibrated ionization chamber (60Co as reference
beam quality for absorbed dose to water) was placed in an
RW3 solid phantom at the nominal treatment distance
(5.5 cm away from the collimator), at the water-equivalent
depth of 1 mm, achieved using the chamber protective cap,
and connected to a PTW Unidos-webline electrometer. Polar-
ity and ion recombination effects, under the approximation of
continuous scanned beams (spill duration of about 1 s being
much longer than the ion chamber charge collection time),
were within the experimental uncertainty (1%); therefore, no
correction factors were applied. The estimated relative stan-
dard uncertainty in absorbed dose determination for this type
of plane-parallel chamber is 2.3%.33
5 Ciocca et al.: Scanning proton ocular beamline 5

A set of 3 9 3 cm2 square mono-energetic uniform fields

(2E+08 particles per spot) was delivered, with a transversal
scanning step of 2 mm, for 16 discrete energies uniformly
distributed within the ocular energy range (62.7–89.8 MeV).
Scanning area and step were selected to maximize dose rate
at the measurement point.
In our standard clinical practice, the calibration factor (C)
for a specific energy (Ei) is defined as the number of particles
(N) per monitor unit (MU),16 that is,

N Dx
Dy
CðEi Þ ¼ ¼ Dmeas.
i ðzref: Þ
(1)
MU
SMC
;
MU i ðzref: Þ

where Dmeas:
i is the absorbed dose measured in the phantom,
SMC
i is the mass stopping power of particles with initial
energy Smeas:
i at the reference depth zref. as calculated using
the FLUKA MC code and (Dx, Dy) is the spot spacing. How-
ever, in this eye-specific scenario, the position of the RS
upstream of the BM, as well as the huge beam scattering
effect and the additional final collimation, lead us to use the
FIG. 2. Experimental setup for secondary neutron equivalent dose measure-
following modified formula: ments to the contralateral eye, using a passive bubble detector taped to the
. phantom anterior surface. Beam comes from right through the patient-speci-
CðEi Þ ¼ Dimeas: ðzref: Þ DMC
i ðzref: Þ; (2) fic brass collimator. Room laser cross at the isocenter is also shown (red
lines).

where DMC
i is the absorbed dose at zref. = 1 mm per unit pri-
mary weight, as predicted by MC code.
2.F. Secondary neutron dose
During any treatment, secondary neutrons are expected
to be generated by beam interaction with the elements
upstream of the patient, primarily the brass collimator
and RS. Neutron equivalent dose to the contralateral eye
was therefore measured with four calibrated passive bub-
ble dosimeters (model BD-PND, medium sensitive range,
Bubble Technology Industries, Chalk River, Canada). The
dosimeters, with a sensitivity of 0.72, 0.76, 0.77 and
0.88 bubbles/lSv, respectively (10% uncertainty), were
placed at the location of the contralateral eye (i.e.,
64 mm lateral from the beam isocenter), fixed to the
head of an anthropomorphic Alderson RANDO phantom
(RSD Radiology Support Devices Inc., Long Beach, CA,
USA).
Each detector was exposed to proton beams under typi-
cal ocular treatment conditions (SOBP with energies in
the range 65 to 86 MeV, RS and final collimator, as
shown in Fig. 2) at a dose of 1.45 Gy (expected number
of produced bubbles around 70–90). In one case, the irra-
diation was delivered at higher dose (4.5 Gy) after
removal of the final collimator, to determine the effect of
this element in neutron production, by comparison with
corresponding results achieved for the collimated beam.
Following detector exposures, bubbles were manually
counted by two independent observers (counting uncer-
tainty around 6%; overall uncertainty in neutron dose
estimation around 12%).
3. RESULTS
3.A. Eyeline setup
As a first step of the eyeline configuration, MC evalua-
tions for the simplified beamline geometry showed that the
proper collimator-to-skin air gap to optimize at the same time
the scattering effect, LP and uniformity at the skin level
resulted in a distance of 5–6 cm. Assuming an isocenter
depth in the patient of about 1.5 cm (eye globe half-thick-
ness), this led us to set a collimator fixed distance of 7 cm
from the isocenter, as for most existing eye-dedicated beam-
lines.10 With shorter air gaps, edge-scattered protons from
aperture generate lateral “horns” on beam profiles, thus wors-
ening the level of dose homogeneity. A further limitation to
shorter air gaps is represented by mechanical interference
with the ETS.
Larger air gaps are instead responsible for the enlargement
of the LP at the field edges. In fact, by varying the collima-
tor-to-skin distance from 0 to 10 cm, the 80%–20% penum-
bra typically increases from about 0.7 to 2 mm.
Concerning the optimal location of the RS along the
beamline, results of MC calculations confirmed that maxi-
mizing the air gap between the beam absorber and the colli-
mator reduces the lateral dose fall-off and improves field
homogeneity. The 80%–20% penumbra roughly lowered by a
factor of 2 when moving the RS, for example, from 30 to
60 cm away from the collimator in the simplified beamline
geometry. Figure 3 gives a schematic representation of the
generation of the calculated collimated profiles. A less pro-
nounced forward-projected dose profile (dashed red lines in
Fig. 3) guaranteed a more extended region of uniformity, this
representing an advantage when a final beam collimation is

Medical Physics, 0 (0), xxxx

6 Ciocca et al.: Scanning proton ocular beamline
FIG. 3. Pictorial view of the lateral beam evolution at different scoring planes
if the range shifter (RS) is placed close to the collimator (black case) or far
from it (red case). Mainly the air gap is responsible for the beam enlarge-
ment, since the range shifter contributes with a comparable scattering power
irrespectively of its location along the beamline.
provided. Therefore, the configuration corresponding to the
most pronounced pencil beam widening appeared the one
characterized by the sharper dose gradient in the penumbra
region of the collimated beam.
As shown in Fig. 1, the RS is placed as close as possi-
ble to the vacuum window to maximize the air gap. In
this eye-dedicated configuration, the Gaussian-shaped spot
size of the uncollimated pencil beam in air at the isocen-
ter (in terms of FWHM) is much larger than in the geom-
etry adopted for conventional treatments (6 cm vs 2 cm,
approximately).
For PBS modality and uncollimated fields, it is well
known that the RS plays a relevant role in terms of actual
beam size and shape at isocenter, therefore the extent of the
LP also depends on how close to the patient the RS can be
positioned.35 However, coupling such an absorber with a
collimator, which is mandatory for ocular treatment, repre-
sents an even more complex issue, as demonstrated by
Wang et al.36 and Gelover et al.,37 who describe the trim-
ming process which a symmetrical pencil beam undergoes
when interacting with a collimator. In general, the air gap
should be chosen to have, at the collimator entrance, a pen-
cil beam size much greater than the aperture diameter. In
case of an extended field, the FS should be compared
against patient-specific collimator aperture, to ensure that
the shifted peak of the trimmed pencil beam remains out-
side or at its edge.36
It should be noted that the idea of increasing the RS to col-
limator air gap by setting an extended treatment distance (i.e.,
shifting the collimator together with the patient well far away
from the nozzle) was excluded, to keep the functionality of
intratreatment patient setup x-ray verification, the existing
PVS being aligned at the standard isocenter.
On machine side, before beam commissioning phase
started, the accelerator physicists worked to find optimal set-
ting of the synchrotron, maximizing proton beam current in
the restricted energy range used in ocular therapy, to shorten
expected treatment time as much as possible.
Medical Physics, 0 (0), xxxx
6
3.B. Depth–dose distributions and SOBP
generation
Figure 4 shows some examples of DDDs acquired with
the Peakfinder system for unscanned, uncollimated mono-
energetic pencil beams, with and without the RS (27-mm
thick RW3), as a reference for MC simulation. Distal fall-off
(90%–10%) did not appreciably depend on the presence or
not of the RS, while increases from 0.8 mm (at 62.7 MeV,
lowest energy) to 1.4 mm (at 89.8 MeV, highest energy set
for ocular treatment) as a function of range. The rWEPLs for
RW3 and PMMA, derived from the analysis of DDDs, are
1.042 and 1.257, respectively. Therefore, with the use of
PMMA thin absorbers, a water-equivalent range resolution of
0.2 mm can be achieved.
Comparison among experimental and MC-calculated
DDDs showed good agreement, with deviations in terms of
BP depth and distal fall-off within 0.15 and 0.02 mm,
respectively, thus confirming the proper choice of the compu-
tational and physics settings adopted for FLUKA simulations.
The whole set of 31 DDDs simulated for eyeline commis-
sioning is shown in Fig. 5.
The distal fall-off (90%–10%) derived from simulated BP
curves for scanning and collimated beams showed again an
increasing trend as a function of the energy, from 1.0 to
1.5 mm. It should be noted that the distal fall-off of the most
distal energy (i.e., the highest one) used to generate any
patient-specific SOBP represents the fall-off of the SOBP
itself.
Figure 6 shows an example of SOBP with a range of
32 mm (i.e., the deepest available one), 17 mm modulation
and 15 mm diameter collimator, calculated by proper weight-
ing of the MC-simulated pristine DDDs.
Comparisons among expected SOBPs and absorbed doses
measured at different depths in water using the ionization
chamber, for ranges between 10 and 32 mm and modulations
between 10 and 17 mm, showed excellent agreement, with
differences in the high-dose region within the experimental
uncertainty, estimated around 0.5%. Deviations found in
the distal fall-off region were also within the uncertainty in
detector measuring depth (0.2 mm) and resolution of the
dose calculation grid (0.1 mm). These findings confirmed
that the general-purpose algorithm proposed by Lomax for
SOBP generation27 well fits even in this context.
3.C. Transversal dose profiles
The irradiation of a large-area EBT3 film in air at the
isocenter, orthogonal to beam direction and in the eyeline
configuration, with an uncollimated 3 9 3 cm2 scanning
field, showed that the proper lateral extent of the field-defin-
ing brass block was around 8 cm. However, further tests at
higher doses, with a collimator having a central aperture of
25 mm diameter, revealed residual low-dose spots outside the
8 9 8 cm2 shielded area. Therefore, to avoid out-of-field
exposure to the patient, a sort of pre-collimator was added
along the beamline [see Fig. 3(e)]: it consisted of 4 cm thick
7 Ciocca et al.: Scanning proton ocular beamline 7

FIG. 4. Examples of integrated depth–dose distributions measured with the Peakfinder system for two energies (81.6 and 89.8 MeV), with (solid lines) and with-
out (dash-dotted lines) the 28.1 mm water-equivalent range shifter (RS). The peak of each curve was normalized to 100%.

FIG. 5. Monte Carlo-calculated depth–dose distributions for the eyeline setup (energy range: 62.7–89.8 MeV; range shifter, 28.1 mm water-equivalent thick,
already included), in terms of dose per unit primary weight.

FIG. 6. Depth–dose distribution for a SOBP with a range of 32 mm and a modulation of 17 mm. Eighteen discrete energy levels were used, from 74.5 to
89.8 MeV. The SOBP calculated starting from Monte Carlo (MC) simulation (solid line) is compared against experimental data (circles) measured with the
Advanced Markus ionization chamber. Error bars are smaller than circle size.

Medical Physics, 0 (0), xxxx

8 Ciocca et al.: Scanning proton ocular beamline 8

FIG. 7. The 80%–20% lateral penumbra as a function of depth in water for five different SOBPs and a collimator of 15 mm diameter. Markers are Monte Carlo-
calculated values, while lines are for visual guide only. R, range (mm); M, modulation (mm).

RW3 slabs, covering an area of approximately 18 9 18 cm2

and having a central 35 mm diameter hole.
Although MC data showed that more than 99% of the pro-
ton beam would be stopped by 6 mm thick brass, for patient
safety 10 mm were prudentially chosen as collimator thick-
ness in the clinical practice.
The analysis of films exposed in the solid phantom
(15 mm depth) to mono-energetic, scanning and collimated
fields did not show any appreciable variation in terms of field
uniformity and LP as a function of scanning area (2 9 2 cm2
vs 3 9 3 cm2) and scan step (2 mm vs 3 mm, this latter rep-
resenting the standard option commonly adopted at CNAO in
proton planning).
In Fig. 7, the 80%–20% LP as a function of depth in
water, calculated from MC simulation for five SOBPs, is FIG. 8. Transversal dose profiles for a SOBP with a range of 20 mm, modu-
lation of 15 mm and 15 mm aperture, at 7 mm (left) and 17 mm depth
shown. In all cases, the expected increase of LP with depth is (right) in an RW3 phantom. Solid lines show Monte Carlo-calculated pro-
evident, due to lateral scattering effect in the material. Range files, while circles are EBT3 experimental data.
being equal (30 mm), LP appears to slightly increase with
modulation, due to the overlapping peripheral dose from each
energy slice. Finally, LP is larger at lower ranges, since scat- respectively. This is due to reduced amount of dose originat-
tering effect is more relevant at lower energies. ing at the field edge, in the outer part of the region where
Figure 8 shows a comparison between measured and MC- uniformity is calculated. Although no corrective action has
calculated lateral dose profiles for a SOBP with a range of been applied so far in case of very large aperture, a modifi-
20 mm, modulation of 15 mm and again the reference cation of the prescribed dose (such as from the prescription
15 mm collimator. An overall good agreement is observed at the standard 100% isodose level to 95%) could be benefi-
for both selected depths (7 and 17 mm), with measured LP cial to compensate underdosage at the tumor periphery,
values differing at most by 0.1 mm from the predicted values. while keeping dose hot spots acceptable (around 63–65 Gy
As expected, FS increased with depth, from 15.6 to 16.0 mm. relative biological effectiveness (RBE)).
As shown in Table I, the analysis of lateral dose profiles Slight increase of penumbra width with field diameter was
at middle-SOBP for different collimators representative of also observed (1.0–1.5 mm and 1.4–1.7 mm for penumbra
ocular clinical practice, showed a non-symmetry never defined at 90–50% and 80–20% isodose levels, respectively).
exceeding 1%, while flatness increased with FS. This latter
was within the optimal level of 3% for field diameters up to 3.D. Calibration of the beam monitor chambers
15 mm (majority of clinical cases), while increased up to
4% and 6% for diameters around 20 (minority of cases) and Figure 9 shows the BM calibration curve used for the
25 mm (very rare cases, involving very large tumors), CNAO eyeline. A representative set of 16 individual C(Ei)

Medical Physics, 0 (0), xxxx

9 Ciocca et al.: Scanning proton ocular beamline 9

TABLE I. Main dosimetric characteristics of lateral dose profiles acquired on EBT3 films irradiated at the depth of middle-SOBP (spread-out-Bragg peak) for
seven different collimators, sorted by increasing field size (see text for details).

Collimator Field size (mm) Flatness (%) Symmetry (%) Penumbra 80%–20% (mm)a Penumbra 90%–50% (mm)a

Circular, 6 mm diameterb 6.2 1.8 0.5 1.4 1.1

Case 1, horizontal axis 9.9 2.0 0.3 1.4 1.2
Case 1, vertical axis 10.3 1.9 0.2 1.4 1.0
Case 2, horizontal axis 14.7 3.0 0.2 1.5 1.2
Circular, 15 mm diameterb 15.8 2.7 0.7 1.4 1.1
Case 4, horizontal axis 16.4 3.5 0.2 1.4 1.2
Case 3, horizontal axis 18.7 3.7 0.6 1.5 1.3
Case 3, vertical axis 21.3 5.4 0.9 1.5 1.2
Case 2, vertical axis 24.4 4.5 0.2 1.7 1.4
Case 4, vertical axis 25.3 6.1 1.0 1.7 1.5
Circular, 25 mm diameterb 26.5 5.9 0.9 1.6 1.4

a
Averaged over each of the two half-profiles.
b
averaged over horizontal and vertical axis profiles.

FIG. 9. Beam monitor calibration curve adopted for the CNAO eyeline. The energy-dependent calibration factor C(Ei) is defined as the number of particles (N)
per monitor unit (MU).

within the whole range of ocular beam energies was
acquired. A third-order polynomial curve was adopted to
fit collected data and was hardcoded in the software
managing the BM system. The curve is given in terms of
number of particles (N) per monitor unit (MU), for con-
sistency with the formalism used for standard treatments.
On its turn, N was derived from absorbed dose to water
measured with the Advanced Markus chamber under refer-
ence conditions [Eq. (2)]. Also in terms of shape, this
eye-specific curve shows similar behavior than the general
one,16 although the individual calibration factors strongly
differ, due to the presence of the RS upstream of BM
chambers, as previously underlined.
3.E. Secondary neutron dose
The average equivalent dose to the contralateral eye, due
to fast neutron contamination of the primary beam, assessed
by three bubble detectors, was 68 lSv/Gy (range 58–79 lSv/
Gy), leading to 3.7 mSv for a typical total dose prescription
of 60 Gy (RBE). Moreover, when removing patient-specific
collimator from the beamline, 24 lSv/Gy (30% of the corre-
sponding value achieved with the collimator inserted) were
recorded. This finding confirmed the expectation about the
brass collimator as the main responsible, together with the
thick plastic RS, for secondary neutron production.
4. DISCUSSION
In modern PBT, PBS has gained more and more interest
worldwide and is rapidly replacing passive scattering as beam
delivery technique, to overcome the following main limita-
tions of scattered beams: need for patient-specific beam mod-
ifying devices, poor capability of dose modulation, lack of
proximal dose conformity.38 However, an exception is repre-
sented by dedicated ocular beamlines, where scattering sys-
tems providing broad fields, although of much smaller size
compared to the scenario characterizing deep tumors, still

Medical Physics, 0 (0), xxxx

10 Ciocca et al.: Scanning proton ocular beamline
remain the state of the art of the technology. Concerning the
proton accelerator for the treatment of ocular tumors, either a
low- or high-energy cyclotron or a synchrotron is normally
used. In the second case, beam energy needs to be strongly
degraded at the exit of the accelerator (from 250 to 70 MeV,
typically); therefore, an increase of beam energy spread and
distal dose fall-off is expected.10 The effects of energy local
degraders in the treatment room on beam transmission, lateral
and distal penumbra have been recently discussed by Gerber-
shagen et al.39 In particular, they reported an increase of dis-
tal fall-off from 1.4 to 2.5 mm with local degrader amount
increase, although leading to higher beam intensities (i.e.,
shorter treatment time). However, the clinical impact of larger
distal penumbra appears limited, since range uncertainty
makes critical organ sparing by beam collimation (i.e., organ
displacement by proper gaze angle selection) preferable than
by distal fall-off.10,39
In our case, the combination of PBS modality and syn-
chrotron-based facility represents a unique scenario within
ocular PBT using single-anterior field approach. Clinical
SOBPs are generated by properly weighting beams
extracted from the accelerator at discrete energy levels from
63 to 90 MeV; our finding in terms of distal fall-off (1.0 to
1.5 mm) is consistent with values between 0.7 and 3 mm
reported for most ocular dedicated beamlines.10 Similarly,
our results of LP width and secondary neutron dose are
similar to those described in the literature10,39,40 and appear
clinically acceptable. In particular, penumbra width is com-
patible with the usual 2.5 mm lateral expansion of the tar-
get defined as block margin in treatment planning. Being
the brass collimator the main responsible for secondary
neutron production, we only measured the equivalent neu-
tron dose at the contralateral eye, very close to the irradi-
ated volume. Indeed, as reported by Slopsema et al.,10 the
neutron dose is considerably higher at this location than at
the chest.
Finally, within our comprehensive quality assurance pro-
gram, our center participated in 2017 into the mailed
dosimetry intercomparison among 15 ocular proton radio-
therapy facilities in Europe and North America. The inter-
comparison was organized by the Polish Institute of Nuclear
Physics (IFJ PAN, Krakow, Poland), in cooperation with the
Working Group 9 of the European Radiation Dosimetry
Group (EURADOS) and the Ocular Proton Therapy Interna-
tional Community (OPTIC), a subcommittee of the Particle
Therapy Cooperative Oncology Group (PTCOG). Five dif-
ferent phantoms with alanine, radio-photoluminescence
(RPL), thermoluminescence (TLD) detectors and EBT3
radiochromic films were exposed in November 2017 to the
dose of 15 Gy in the eye setup (collimator diameter equal to
25 mm) and shipped back to Poland for readout and analy-
sis. Preliminary results of the intercomparison were available
in May 2018 and in our case showed average agreement
within 3% against reference doses (J. Swakon, IFJ PAN,
Krakow, Poland, personal communication to all participating
centers). However, final results and data analysis are still
missing and not yet published.
Medical Physics, 0 (0), xxxx
10
5. CONCLUSIONS
This study showed that maximizing the air gap between
RS and patient is beneficial in terms of improved lateral beam
penumbra for collimated, scanning proton beams. Optimal
distance of 5.5 cm between patient-specific collimator and
eye surface was found.
The presence of the fixed RS upstream of the BM cham-
bers required a dedicated calibration procedure, different
from the standard one used for non-ocular treatments.
Our adapted proton horizontal beamline satisfied the
requirements for safe and proper treatment of intraocular
tumors at CNAO.
For the dose fractionation scheme currently adopted in our
center for uveal melanoma (60 Gy (RBE) in four daily frac-
tions, using a fixed RBE value of 1.1), beam delivery time is
around 3 min per session. Starting from August 2016, more
than 100 patients so far successfully completed their ocular
treatment (December 2018).
Finally, the strategy of combining MC simulations and
experimental data acquisition in supporting our beamline
design and ocular treatment start-up phase appeared highly
successful, as already reported for other scenarios as well as
dedicated eyelines.39–43
CONFLICTS OF INTEREST
The authors have no conflicts of interest to disclose.
a)
Author to whom correspondence should be addressed. Electronic mail:
edoardo.mastella@cnao.it; Telephone: +39 0382 078509.
REFERENCES
1. Hrbacek J, Mishra KK, Kacperek A, et al. Practice patterns analysis of
ocular proton therapy centers: the international OPTIC survey. Int J
Radiat Oncol Biol Phys. 2016;95:336–343.
2. Nathan P, Cohen V, Coupland S, et al. Uveal melanoma UK national
guidelines. Eur J Cancer. 2015;51:2404–2412.
3. Egger E, Schalenbourg A, Zografos L, et al. Maximizing local tumor
control and survival after proton beam radiotherapy of uveal melanoma.
Int J Radiat Oncol Biol Phys. 2001;51:138–147.
4. Polishchuk AL, Mishra KK, Weinberg V, et al. Temporal evolution and
dose-volume histogram predictors of visual acuity after proton beam
radiation therapy of uveal melanoma. Int J Radiat Oncol Biol Phys.
2017;97:91–97.
5. Damato B, Kacperek A, Chopra M, Campbell IR, Errington RD. Proton
beam radiotherapy of choroidal melanoma: the Liverpool-Clatterbridge
experience. Int J Radiat Oncol Biol Phys. 2005;62:1405–1411.
6. Damato B, Kacperek A, Errington D, Heimann H. Proton beam radio-
therapy of uveal melanoma. Saudi J Ophthalmol. 2013;27:151–157.
7. Kacperek A. Protontherapy of eye tumours in the UK: a review of treat-
ment at Clatterbridge. Appl Radiat Isot. 2009;67:378–386.
8. Sikuade MJ, Salvi S, Rundle PA, Errington DG, Kacperek A, Rennie
IG. Outcomes of treatment with stereotactic radiosurgery or proton
beam therapy for choroidal melanoma. Eye (Lond). 2015;29:1194–1198.
9. Weber DC, Abrunhosa-Branquinho A, Bolsi A, et al. Profile of Euro-
pean proton and carbon ion therapy centers assessed by the EORTC
facility questionnaire. Radiother Oncol. 2017;124:185–189.
10. Slopsema RL, Mamalui M, Zhao T, Yeung D, Malyapa R, Li Z. Dosi-
metric properties of a proton beamline dedicated to the treatment of ocu-
lar disease. Med Phys. 2014;41:011707-1–011707-11.
11 Ciocca et al.: Scanning proton ocular beamline
11. Hartsell WF, Kapur R, Hartsell SO, et al. Feasibility of proton beam
therapy for ocular melanoma using a novel 3D treatment planning tech-
nique. Int J Radiat Oncol Biol Phys. 2016;95:353–359.
12. Mosci C, Mosci S, Barla A, Squarcia S, Chauvel P, Iborra N. Proton
beam radiotherapy of uveal melanoma: Italian patients treated in Nice,
France. Eur J Ophthalmol. 2009;19:654–660.
13. Spatola C, Privitera G, Raffaele L, et al. Clinical application of proton
beams in the treatment of uveal melanoma: the first therapies carried out
in Italy and preliminary results (CATANA Project). Tumori. 2003;89:
502–509.
14. Cuttone G, Cirrone GAP, Pittera S, et al. CATANA protontherapy facil-
ity: the state of art of clinical and dosimetric experience. Eur Phys J
Plus. 2011;126:1–7.
15. Rossi S. The national centre for oncological hadrontherapy (CNAO): sta-
tus and perspectives. Phys Med. 2015;31:333–351.
16. Mirandola A, Molinelli S, Vilches Freixas G, et al. Dosimetric commis-
sioning and quality assurance of scanned ion beams at the Italian
National Center for Oncological Hadrontherapy. Med Phys. 2015;42:
5287–5300.
17. Giordanengo S, Garella MA, Marchetto F, et al. The CNAO dose deliv-
ery system for modulated scanning ion beam radiotherapy. Med Phys.
2015;42:263–275.
18. Jelen U, Bubula ME, Ammazzalorso F, Engenhart-Cabillic R, Weber U,
Wittig A. Dosimetric impact of reduced nozzle-to-isocenter distance in
intensity-modulated proton therapy of intracranial tumors in combined
proton-carbon fixed-nozzle treatment facilities. Radiat Oncol. 2013;
8:218.
19. Pedroni E, Scheib S, B€ohringer T, et al. Experimental characterization
and physical modelling of the dose distribution of scanned proton pencil
beams. Phys Med Biol. 2005;50:541–561.
20. Piersimoni P, Rimoldi A, Riccardi C, Pirola M, Molinelli S, Ciocca M.
Optimization of a general-purpose, actively scanned proton beamline for
ocular treatments: Geant4 simulations. J Appl Clin Med Phys. 2015;
16:5227.
21. Via R, Fassi A, Fattori G, et al. Optical eye tracking system for real-time
noninvasive tumor localization in external beam radiotherapy. Med Phys.
2015;42:2194–2202.
22. Ferrari A, Sala PR, Fasso A, Ranft J. FLUKA: a multi-particle transport
code INFN-TC-05–11, SLAC-R-773, CERN-2005-10. 2005.
23. B€ohlen TT, Bauer J, Dosanjh M, et al. A Monte Carlo-based treatment-
planning tool for ion beam therapy. J Radiat Res. 2013;54:i77–i81.
24. Molinelli S, Mairani A, Mirandola A, et al. Dosimetric accuracy assess-
ment of a treatment plan verification system for scanned proton beam
radiotherapy: one-year experimental results and Monte Carlo analysis of
the involved uncertainties. Phys Med Biol. 2013;58:3837–3847.
25. Mairani A, B€ ohlen TT, Schiavi A, et al. A Monte Carlo-based treatment
planning tool for proton therapy. Phys Med Biol. 2013;58:2471–2490.
26. Parodi K, Mairani A, Brons S, et al. Monte Carlo simulations to sup-
port start-up and treatment planning of scanned proton and carbon ion
therapy at a synchrotron-based facility. Phys Med Biol. 2012;57:3759–
3784.
27. Lomax A. Intensity modulation methods for proton radiotherapy. Phys
Med Biol. 1999;44:185–205.
Medical Physics, 0 (0), xxxx
11
28. McAuley GA, Teran AV, Slater JD, Slater JM, Wroe AJ. Evaluation of
the dosimetric properties of a diode detector for small field proton radio-
surgery. J Appl Clin Med Phys. 2015;16:51–64.
29. Raffaele L, Ciocca M, Mirandola A, Spampinato S, Cirrone GAP. Dosi-
metric characterization of a silicon diode detector in cyclotron-based
passively scattered and synchrotron-based scanning clinical proton
beams. Nucl Instrum Methods Phys Res A. 2018;891:125–132.
30. Marinelli M, Pompili F, Prestopino G, et al. Dosimetric characterization
of a synthetic single crystal diamond detector in a clinical 62 MeV ocu-
lar therapy proton beam. Nucl Instrum Methods Phys Res A.
2014;767:310–317.
31. Marinelli M, Prestopino G, Verona C, et al. Dosimetric characterization
of a microDiamond detector in clinical scanned carbon ion beams. Med
Phys. 2015;42:2085–2093.
32. Rossomme S, Marinelli M, Verona-Rinati G, et al. Response of syn-
thetic diamond detectors in proton, carbon and oxygen ion beams. Med
Phys. 2017;44:5445–5449.
33. International Atomic Energy Agency (IAEA). Absorbed Dose Determi-
nation in External Beam Radiotherapy: An International Code of Prac-
tice for Dosimetry Based on Standards of Absorbed Dose to Water.
Technical Reports Series No. 398. Vienna: IAEA; 2000.
34. International Commission on Radiation Units and Measurements
(ICRU). Prescribing, recording, and reporting proton-beam therapy.
Report No. 78. J ICRU. 2007;7:29–48.
35. Lomax AJ, B€ohringer T, Bolsi A, et al. Treatment planning and verifica-
tion of proton therapy using spot scanning: initial experiences. Med
Phys. 2004;31:3150–3157.
36. Wang D, Smith BR, Gelover E, Flynn RT, Hyer DE. A method to select
aperture margin in collimated spot scanning proton therapy. Phys Med
Biol. 2015;60:N109–N119.
37. Gelover E, Wang D, Hill PM, et al. A method for modeling laterally
asymmetric proton beamlets resulting from collimation. Med Phys.
2015;42:1321–1334.
38. Engelsman M, Schwarz M, Dong L. Physics controversies in proton
therapy. Semin Radiat Oncol. 2013;23:88–96.
39. Gerbershagen A, Hrbacek J, Ijpes D, Schippers JM. Transmission
improvement options via local energy degradation at a cyclotron driven
ocular tumor treatment facility. Modern Phys Lett A. 2017;32:1740016-
1–1740016-13.
40. Titt U, Suzuki K, Li Y, Sahoo N, Gillin MT, Zhu XR. Technical note:
dosimetric characteristics of the ocular beam line and commissioning
data for an ocular proton therapy planning system at the Proton Therapy
Center Houston. Med Phys. 2017;44:6661–6671.
41. Magro G, Molinelli S, Mairani A, et al. Dosimetric accuracy of a treat-
ment planning system for actively scanned proton beams and small target
volumes: Monte Carlo and experimental validation. Phys Med Biol.
2015;60:6865–6880.
42. Herault J, Iborra N, Serrano B, Chauvel P. Monte Carlo simulation of a
protontherapy platform devoted to ocular melanoma. Med Phys. 2005;
32:910–919.
43. Baker CR, Quine TE, Brunt JN, Kacperek A. Monte Carlo simulation
and polymer gel dosimetry of 60 MeV clinical proton beams for the
treatment of ocular tumours. Appl Radiat Isot. 2009;67:402–405.