See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/334497706

Experience in commissioning The Halcyon linac

Article  in  Medical Physics · July 2019

DOI: 10.1002/mp.13723

CITATIONS READS

18 2,742

17 authors, including:

Yuting Li Song Gao

University of Texas MD Anderson Cancer Center University of Texas MD Anderson Cancer Center
20 PUBLICATIONS   220 CITATIONS    73 PUBLICATIONS   759 CITATIONS   

SEE PROFILE SEE PROFILE

Laurence E Court Lei Dong

University of Texas MD Anderson Cancer Center University of Pennsylvania
405 PUBLICATIONS   6,096 CITATIONS    484 PUBLICATIONS   14,134 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

“Archimedean TEI ATHENS EPEAEK II” with title: «Automatic Diagnosis System of Thyroid Nodules Cancer "Department of Medical Instruments Technology T.E.I. Athens.
View project

Ultra-high-dose rate (FLASH) proton therapy View project

All content following this page was uploaded by Lei Dong on 13 January 2020.

The user has requested enhancement of the downloaded file.

Experience in commissioning the halcyon linac
Tucker Netherton, Yuting Li, Song Gao, Ann Klopp, Peter Balter, and Laurence E Court
Department of Radiation Physics, University of Texas MD Anderson Cancer Center, Houston, TX 77030, USA
Ryan Scheuermann, Chris Kennedy, Lei Dong, James Metz, and Dimitris Mihailidisa)
Perelman Center for Advanced Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA
Clifton Ling, Mu Young Lee, Magdalena Constantin, Stephen Thompson, Juha Kauppinen,
and Pekka Uusitalo
Varian Medical Systems Inc, Palo Alto, CA 94304, USA

(Received 15 March 2019; revised 10 June 2019; accepted for publication 1 July 2019;
published 27 August 2019)
Purpose: This manuscript describes the experience of two institutions in commissioning the new
HalcyonTM platform. Its purpose is to: (a) validate the pre-defined beam data, (b) compare relevant
commissioning data acquired independently by two separate institutions, and (c) report on any signif-
icant differences in commissioning between the Halcyon linear accelerator and other medical linear
accelerators.
Methods: Extensive beam measurements, testing of mechanical and imaging systems, including the
multi-leaf collimator (MLC), were performed at the two institutions independently. The results were
compared with published recommendations as well. When changes in standard practice were necessi-
tated by the design of the new system, the efficacy of such changes was evaluated as compared to
published approaches (guidelines or vendor documentation).
Results: Given the proper choice of detectors, good agreement was found between the respective
experimental data and the treatment planning system calculations, and between independent measure-
ments by the two institutions. MLC testing, MV imaging, and mechanical system showed unique
characteristics that are different from the traditional C-arm linacs. Although the same methodologies
and physics equipment can generally be used for commissioning the Halcyon, some adaptation of
previous practices and development of new methods were also necessary.
Conclusions: We have shown that the vendor pre-loaded data agree well with the independent mea-
sured ones during the commission process. This verifies that a data validation instead of a full-data
commissioning process may be a more efficient approach for the Halcyon. Measurement results
could be used as a reference for future Halcyon users. © 2019 American Association of Physicists in
Medicine [https://doi.org/10.1002/mp.13723]

Key words: beam data verification, commissioning, Halcyon, TPS verification.

1. INTRODUCTION version (Halcyon Version 1.1) and repeated commissioning

This study describes the combined experiences of two institu-
tions in commissioning a recently released radiotherapy treat-
ment device which has some significant differences
compared with conventional C-arm Linacs. Prior to its clini-
cal release, the vendor (Varian Medical Systems, Palo Alto,
CA) contracted and worked with two academic institutions,
the University of Texas MD Anderson Cancer Center
(MDACC) and the Hospital of University of Pennsylvania
(UPENN), for independent clinical validation of a pre-release
version of the new treatment machine and pre-release version
of the Eclipse treatment planning software. The research was
conducted independently without any communication
between the two institutions until two months prior to the
public announcement of the new system. Then, all commis-
sioning data gathered by the two institutions were compared.
The new linac was named as HalcyonTM Version 1.0 (Varian
Medical Systems, Palo Alto, CA). Later, UPENN replaced
the pre-release version of Halcyon system with the production
measurements. The linac at MDACC was a non-clinical,
research only machine. University of Pennsylvania success-
fully treated the first patient on September 14, 2017 using the
production version and has treated over 250 patients to date.
The goal of this work is to: (a) validate the pre-defined beam
data, (b) compare relevant commissioning data acquired from
two separate institutions, and (c) report on any significant dif-
ferences in commissioning between the Halcyon linear accel-
erator and other medical linear accelerators.
2. BACKGROUND
The main features of this new treatment device are sum-
marized in Table I. It has an enclosed, ring-mounted gantry
and a 6 MV flattening filter free (FFF) beam. This beam is
used for both treatment and imaging for Halcyon version 1.0.
Future versions of Halcyon (Halcyon 2) will include a kV
imaging system. Because the gantry is enclosed, it is allowed
to rotate significantly faster than on conventional C-arm

4304 Med. Phys. 46 (10), October 2019 0094-2405/2019/46(10)/4304/10 © 2019 American Association of Physicists in Medicine 4304
4305 Netherton et al.: Commissioning the Halcyon
TABLE I. Features of the Halcyon treatment device (with Eclipse treatment
planning system).
Halcyon Feature Description
Beam model Pre-defined
Ring mounted gantry 15 s rotation time
Enclosed gantry bore 100 cm diameter
Treatment and imaging beam 6 MV FFF
Nominal dose rate 800 MU/min
Dual layer, stacked MLC design Dynamic MLC, 5 cm/s leaf speed
Maximum field size 28 9 28 cm2
Varian IGRT Couch Coplanar with mounted camera
Alignment at isocenter IGRT only, no light field, no ODI
treatment devices – with a maximum rotation speed of 360°
in 15 s. The Halcyon 1.0 linac only permits the delivery of
dynamic multi-leaf collimator (MLC) plans in clinical mode
(field-in-field, IMRT, or RapidArcâ, or irregular surface
compensator), and has a dual-layered, stacked MLC that pro-
jects to a maximum field size of 28 cm2 9 28 cm2 at isocen-
ter. Each leaf is 1 cm wide with 58 upper and 56 lower MLCs
for a total of 114 leaves. The top layer is displaced by 0.5 cm
laterally relative to the lower layer1 to reduce linter-leaf leak-
age. The leaves have a maximum speed of 5.0 cm/s, twice
that of current Varian MLCs. There are no moving collimat-
ing jaws because the dual-stack MLC design is designed to
provide adequate shielding and allows for “per-leaf tracking”
as opposed to the jaw tracking feature in other C-arm linacs.
In regard to the typical workflow on the Halcyon, the
patient (or physics QA equipment) is first positioned using
lasers outside of the Halcyon gantry housing. The couch is
then translated into the bore by a specific distance, moving
the patient to the treatment position. MV images are then
acquired to verify that the patient is accurately positioned for
treatment. As there is neither light field, optical-distance-indi-
cator, nor mechanical distance measurement device, accurate
patient set-up relies largely on MV imaging. In-line of the
radiation source, on the opposite side of the gantry is an elec-
tronic portal imaging device (EPID) positioned in front of a
beam stopper. In the versions that we tested (v1.0 and 1.1), all
treatments were image-guided, using either MV orthogonal
projections or MV CBCT. For each image method, there is a
choice of “low-dose” or “high-quality” mode. Since, the dose
from this imaging beam is included in the treatment plan, the
choice of the imaging technique must be made before treat-
ment planning commences.
Varian provides a representative beam data set (RBD),
consisting of percentage depth doses (PDDs), profiles and
output factors. This RBD set is used by Varian to create the
pre-configured Halcyon optimization and dose calculation
models (PO & AAA) in Eclipse 15.1.1. treatment planning
system (TPS)1.
This study describes the pre-clinical testing of the Halcyon
1.0 (two units, one at each institution), and the subsequent
testing of a clinical version of the device at UPENN. The test-
ing is exhaustive and largely based on relevant American
Medical Physics, 46 (10), October 2019
4305
Association of Physicists in Medicine (AAPM) and IAEA
documents2–11. A large portion of the test results (common to
both Halcyon and other linacs) are available via an electronic
supplement. The characteristics of the clinical (1.1) and pre-
clinical (1.0) units are very similar, with the only exception
being that the MLC was redesigned for the clinical (1.1) unit
(tongue-and-groove vs no tongue-and-grove). Furthermore,
because the physical design of this treatment device is differ-
ent from previous Varian linacs, we have included some of
our specific experience in the setup and use of QA devices
for this specific unit.
3. MATERIALS AND METHODS
3.A. Independent beam data validation
The performance and stability of the Varian TrueBeam is
well documented across multiple institutions.12,13 The cre-
ation of a RBD set is useful as (a) a robust method of ensur-
ing general consistency across machines of the same design
and (b) a reliable second check for institutions that perform
their own modeling. It has yet to be shown how the perfor-
mance of the Halcyon varies from institution to institution
while using its own unique, and vendor-supplied RBD data.
In the present study, beam data (beam profiles, percent-depth
dose, output factors, and off-axis ratios) were measured and
compared. In addition, the measured results were compared
to the vendor-supplied beam measurements.
3.A.1. Percent depth dose measurements
The percent depth dose (PDD) of 6 MV FFF photon
beams for different field sizes using CC13 ionization cham-
bers (IBA) and diodes (SunNuclear) were measured in the
Blue2 water tank (IBA). Because the diameter of the treat-
ment bore is 100 cm, some other scanning water tanks may
be too large to use for measurements on the Halcyon. In addi-
tion, the water tank cannot be aligned with lasers, light field,
or mechanical measuring device at isocenter. The user must
first align the water tank to lasers at an external setup posi-
tion, then shift the tank to isocenter for subsequent alignment
using image guidance. When the desired source to water sur-
face distance (SSD) is 100 cm, the air-to-water interface can
be easily visualized by taking images with the gantry at either
90 or 270 degrees in service mode. The 100 cm SSD is
achieved by adjusting the table height until the air-water inter-
face is aligned with the crosshair overlay of the isocenter
location in the service mode images. However, when the
desired SSD is not 100 cm, measuring the appropriate table
shifts from images is complicated due to beam divergence
and the blurring of the air-to-water interface (Note: The Var-
ian user manual thoroughly describes this setup procedure for
all users, we recommend that this resource be used). To
achieve an SSD of 90 cm, the gantry is positioned at 84.3
degrees, and the table height is adjusted until the distance
between the isocenter projection in the image and the sharp
air-to-water interface is 10.05 cm. This distance is measured
4306 Netherton et al.: Commissioning the Halcyon
using the measuring tools available in service mode and
relies on a trigonometric relationship indicated in the user
manual. Further alignment of the ion chamber’s measurement
volume to machine isocenter was achieved using the Blue2
Phantom central axis measurement utility, and verified with
an anterior-posterior EPID image using ~ 25–50 MU. The
OmniPro –Accept software (IBA) was used to analyze the
resulting scanning data. Percent depth dose curves for field
sizes: 2 9 2, 4 9 4, 6 9 6, 8 9 8, 10 9 10, 20 9 20, and
28 9 28 cm2 were compared on a point-by-point basis
against the RBD from each institution for depths beyond
dmax. The measurement devices used were CC13 and diodes
(<6 9 cm2) at MDACC, CC04 at UPENN, and CC13 and
diode (<8 9 8 cm2) at Varian for the RBD, respectively.
Comparison against TPS PDDs were completed in section G
(beam model verification).
3.A.2. Dose in the buildup region
For both institutions, dose in the buildup region was mea-
sured in a solid water phantom using parallel-plate chamber
(Table II). The source-to-chamber distance was 100 cm and
dmax was searched for different field sizes. Various thick-
nesses of the solid water blocks were used for depth dose
measurements. Measurements at different depths were nor-
malized to the measured dose at dmax. Since the polarity
effect of the Advanced Markus parallel-plate chamber is very
large in the buildup region (MDACC), measurements were
done in both +300 V and 300 V, and the results averaged.
3.A.3. Beam profiles
Beam profiles were measured at different depths along
in-plane and cross-plane directions at SSD of 90 cm for a
TABLE II. Institutional measurement device Table.
Type Model
Cylindrical ion chambers IBA CC04
IBA CC04
IBA CC13
IBA CC13
IBA CC13
Exradin A12
PTW TN30013
Diodes Sun Nuclear EDGE
Sun Nuclear EDGE
Array Detectors Sun Nuclear QailyQA3
Sun Nuclear IC Profiler
Sun Nuclear Map Check 2
Sun Nuclear Arc Check
Portal Dosimetry
Plane Parallel PTW 23343 Markus
PTW Advanced Markus
PPC40
DLG, Dosimetric leaf gap; MDACC, MD Anderson cancer center; UPENN, university of pennsylvania.
Medical Physics, 46 (10), October 2019
4306
range of field sizes: 2 9 2, 4 9 4, 6 9 6, 8 9 8, 10 9 10,
20 9 20, and 28 9 28 cm2. Measurements were carried
out at five different depths: 1.3 cm (dmax), 5 cm, 10 cm,
20 cm, and 30 cm. The beam profiles were also measured
at five different depths along the diagonal direction of the
largest field size 28 9 28 cm2 at SSD of 90 cm. The
choices of detector for beam profiles were CC13 for FS
>6 9 6, diode for FS <=6 9 6 (MDACC), and CC04 for
all field sizes (UPENN). Representative beam data set was
scanned with a CC13 for FS >4 9 4 and diode for FS<=
4 9 4. For the measured beam profiles, the data were first
smoothed then corrected for central axis discrepancies and
made symmetrical. The beam profiles were then renormal-
ized to 100% by the values at central axis. A point-by-point
comparison between institutional and RBD was used to
evaluate beam profiles. High dose regions, penumbra, and
low dose tails were individually analyzed using point-by-
point and distance to agreement at every field size for
depths 1.3 cm and 10.0 cm (cross-plane and in-plane). To
define each region of the profile, inflection points were
found by plotting the first derivatives of the beam profiles
for each field size (Fig. 1). The full-width-half-maximum of
each first derivative curve was used to define penumbra,
low-dose tail, and high dose regions. Since the 80–20%
definition of penumbra is only directly applicable to flat-
tened profiles, these regions must be explicitly defined in
order for MPPG5a guidelines to be applied. These regions
were analytically determined for the RBD and for the TPS
data and were used in sections 4.A and 4.D.
Flatness and symmetry were calculated from in-plane (ra-
dial) and cross-plane (transverse) scans for 2 9 2 cm2 to
28 9 28 cm2 field sizes at an SSD of 90 cm and different
depths in water. Although flatness is recommended by
AAPM for flattened beams, flatness is not an appropriate
Institution Tests Performed
UPENN All PDD, all beam profiles
MDACC ≤6x6 cm2 output factors, phantom imaging dose
MDACC ≥6x6 cm2 output factors, ≥4x4 cm2 PDD and beam profiles
UPENN TG 119 testing
Varian ≥6 9 6 cm2 PDD and beam profiles, all FS output factors
UPENN TG51
MDACC TG51, monthly output standard, DLG and transmission
MDACC ≤4 9 4 cm2 PDD and ≤6 9 6 beam profiles
Varian <6 9 6 cm2 PDD
MDACC Daily output standard
MDACC Output vs gantry, flattness and symmetry
UPENN IMRT patient specific QA
MDACC, UPENN RapidArc patient specific QA
MDACC, UPENN IMRT and RapidArc patient specific QA
UPENN Surface dose
MDACC Surface dose
UPENN DLG and transmission
4307 Netherton et al.: Commissioning the Halcyon
FIG. 1. Methodology to determine dose regions for a FFF beam. Field size = 10 9 10 cm2, depth = 10 cm. Red curve = TPS, Black curve = RBD using a
CC13 ionization chamber. Blue region = low-dose tails, White region = penumbra, Red region = high-dose region. Here, regions are based on TPS (red). FFF,
flattening filter free; RBD, representative beam data set; TPS, treatment planning system. [Color figure can be viewed at wileyonlinelibrary.com]
metric for this FFF beam. Instead, off axis ratio (including
diagonal normalized flatness) is the appropriate metric to
characterize the open field beam quality.
3.A.4. Output factor measurements
To verify Halcyon RBD set, output factors were measured
with the ionization chamber set at a depth of 5.0 cm and SSD
at 95 cm in the Blue2 water phantom. The SSD was set using
the procedure in the Varian “Instructions for Use” manual.
The measured results for different field sizes were normalized
to that of a 10 9 10 cm2 field to yield the output factors. A
CC04 ion chamber was used for field sizes less than
6 9 6 cm2, CC13 ion chamber was used for field sizes
greater than 6 9 6 cm2 (MDACC). The RBD used a CC13
ion chamber for all field sizes.
3.B. Mechanical system validation against
specifications
Due to the similarities in the method of treatment deliv-
ery, Halcyon mechanical testing requirements are generally
consistent with C-arm linacs.3 While Halcyon’s bore
geometry eliminates the need to test physical indicators of
the treatment field (optical distance indicator, light field,
Medical Physics, 46 (10), October 2019
4307
etc.), it presents challenges for testing by restricting access
to machine components and preventing direct measure-
ment. Nearly all information regarding mechanical perfor-
mance relative to treatment isocenter must be derived from
irradiation of film, EPID imaging, or other phantom-based
radiation measurement.
An additional consideration when evaluating a QA pro-
gram for Halcyon is the implementation of the machine
performance check procedure (MPC),14 which must be exe-
cuted daily prior to treatment, as part of the Halcyon
workflow. The MPC utilizes a drum phantom with imbed-
ded fiducials mounted to a specific position on the treat-
ment couch. The MPC automatically sets various machine
parameters and acquires 33 radiation images with the
EPID. Based on these measurements, the MPC software
performs analysis and verifies that the machine’s dosimet-
ric and mechanical performance are within their respective
tolerances. Another version of the MPC has been shown
previously to provide excellent results and long-term stabil-
ity for QA of TrueBeam.14 We have also carried out mea-
surements, in which errors were introduced intentionally,
to test the ability of the MPC in detecting the errors.15
The methods described below provide an independent
check of the necessary QA procedures to ensure accurate
treatment delivery.
4308 Netherton et al.: Commissioning the Halcyon
3.B.1. Lasers (virtual-iso) versus machine treatment
isocenter
All lasers - lateral (left and right), and vertical (top) - were
checked against the virtual treatment isocenter. The Winston-
Lutz (WL) phantom was used for these tests. The tests were
done as follows: (a) Align the center of the ball with the
lasers, (b) move couch into the treatment isocenter position
by applying the machine’s automated shift from virtual to
treatment isocenter, and record the actual couch coordinates.
(c) Align the center of WL ball to the treatment isocenter
(digital crosshair) by adjusting the couch position using
orthogonal images—final couch coordinates were compared
to that obtained from the “load” operation.
3.B.2. Gantry and collimator angle indicators
As the Halcyon’s gantry is not directly accessible to its user,
(no light field), new methods need to be devised to evaluate
the accuracy of the gantry angle indicator. Two methods were
evaluated: (a) gantry spoke shot referenced to bubble level
indicator, and (b) observation of fiducial movement in portal
images at various couch vertical positions. In the first test, a
horizontal line was marked on a radiochromic film (EBT3)
with the aid of a bubble level indicator, prior to star shot irradi-
ation at various gantry angles. The film was digitized, and the
angle of the spokes relative to the bubble reference line were
evaluated using the measuring tools in the ImageJ software.16
The measured gantry angles were then compared to the corre-
sponding values of gantry positions indicators.
In the second test, where the gantry angle is evaluated based
on fiducial image projections under couch motion, the portal
imager, a WL pointer phantom, variable couch positions, and
in-house software were used. This test’s goal is to address the
accuracy of the angle given by the digital gantry angle indica-
tor. Since this test was found to be sensitive to setup error, the
experiment was repeated with intentional table setup perturba-
tions of 1 mm in the direction sensitive to each measurement.
Cardinal gantry angle was calculated using the portal images
of a WL pointer as a function of table height.
Collimator angle indicator accuracy was assessed by ana-
lyzing EPID images of long narrow fields defined by MLCs
acquired with the collimator digitally set at 0° and 90°. An
in-house (MDACC) machine QA software was used for ana-
lyzing the images.
3.C. Imaging systems validation and performance
Evaluation of the imaging performance of the Halcyon
closely followed standard practice (as defined in MPPG2a7)
to establish baseline values, as expected for any medical lin-
ear accelerator. However, the imaging dose evaluation
revealed some interesting features for Halcyon and its testing
is described below.
A CT of an anthropomorphic thorax phantom (CIRS) was
imported into the Eclipse TPS and the imaging doses calcu-
lated for 11 different points for each of the imaging techniques
Medical Physics, 46 (10), October 2019
4308
available with the Halcyon (including different isocenter posi-
tion and field size combinations). The phantom was then posi-
tioned at the Halcyon and the imaging doses measured for the
same conditions using a cross-calibrated CC04 ion chamber.
The experimentally measured doses were compared with
those calculated by the Eclipse. For more details on our
exhaustive studies into the imaging dose with the Halcyon,
readers are referred to Li et al.17 and Mihailidis et al.18.
3.D. Beam model verification, end-to-end testing,
and treatment plan validation
As part of the overall dosimetric performance validation of
the Halcyon, various aspects of the MPPG5a testing were
performed to ensure that the pre-defined beam model calcula-
tions matched commissioning measurements. The equipment
used for validation was taken from the recommendations of
Tables I and II of MPPG5a.6 Four categories of tests included
in report are TPS model comparison, basic photon beam vali-
dation, heterogeneous photon beam validation, and RA/
IMRT validation.
For TPS model comparisons the tests performed were: (a)
dose in test plan vs clinical calibration condition, and (b) dose dis-
tribution calculated in planning system vs. commissioning data
(PDD, off-axis output factors). In a situation where the user can-
not obtain a set of measured beam data to compare with the TPS,
the RBD can be requested from the vendor. For PDD and beam
profile data comparisons against the TPS, MPPG5a recom-
mended distance to agreement for penumbra (3 mm), high dose
region point dose agreement (2.0%), and low-dose profile tails
point dose agreement (3%) criteria were used. Dose region speci-
fications for these flattening-filter-free beam profiles were calcu-
lated as mentioned in section B (independent beam validation).
The following basic beam validation tests were performed:
(a) PDD and profile comparison of small square MLC shaped
fields, (b) PDD and profile comparison of asymmetric MLC
shaped fields, (c) point dose and profile comparison of large
field with extensive MLC blocking, (d) point dose and profile
comparison of half-beam block at minimally anticipated SSD,
and (e) point dose comparison of oblique incident beam.
Exported doses from the Eclipse TPS used a 0.25 cm grid size.
The following tests were performed for heterogeneity cor-
rection validation: (a) Point dose comparison (with AAA for
treatment dose) using an anthropomorphic phantom (CIRS)
and (b) point dose comparison (with modified AAA used for
imaging dose) using an anthropomorphic phantom (CIRS).
Dosimetric leaf gap (DLG) measurements were performed
for the Halcyon MLC system prior to any dynamic delivery
testing and was performed by using the procedure and dicom
files provided by the manufacturer. Finally, for IMRT/RA dose
validation the tests performed were: (a) AAPM TG-119 testing,
(b) clinical tests (end-to-end), and (c) external review using
IMRT and RA (IROC end-to-end). Output factor and PDD ver-
ification for very small MLC field sizes were also performed
in the above sections, since all fields on the Halcyon are
defined by the MLC. Point measurements described in AAPM
TG 1198 were performed using an IBA CC13 cylindrical ion
4309 Netherton et al.: Commissioning the Halcyon 4309

chamber in a solid water phantom. All plans were generated
using both IMRT and RA delivery techniques. Measured dose
values were compared against the average dose calculated in
Eclipse within the chamber collecting volume.
For the external review process, we created plans and irra-
diated the anthropomorphic head (RA plan) and thorax phan-
tom (IMRT plan) from IROC-Houston, following the
standard guidelines. Patient-specific QA measurements were
also performed at UPENN for all plans using either the Sun
Nuclear Corporation MapCHECKâ 2 (IMRT), ArcCHECKâ
(RA) diode arrays, or portal dosimetry. Distance to agreement
(DTA) analysis was performed on all measurements using
two methods: 1) 3%/3mm, global normalization, 10% thresh-
old, and 2) 2%/2 mm, local normalization, 10% threshold.
4. RESULTS
4.A. Independent Beam Data validation
4.A.1. Percent depth dose measurements
For all field sizes measured (2 9 2 cm2 to 28928 cm2),
PDD curves from both institutions matched the vendor pro-
vided RBD within 1.0% (step size = 1 mm). For field sizes
greater than 2 9 2 cm2, this maximum difference was 0.6%.
The maximum point difference between the RBD and mea-
sured data is as follows for various field sizes (cm2) at each
institution and is featured in Table III: (for MDACC and
UPENN).
4.A.2. Buildup region dose
Dose in the buildup region was measured in a solid block
phantom using a parallel-plate chamber (both institutions).
The source-to-chamber distance was 100 cm and dmax was
searched for different field sizes. Various thicknesses of the
blocks were used for surface depth dose measurements. Mea-
surements at different depths were normalized to the mea-
sured dose at dmax.
TABLE III. Maximum percent difference, measured data vs RBD.
For a range of depths (1–15 mm) and square field sizes
(6 9 6, 10 9 10, and 20 9 20 cm2), ratios of surface doses
between institutions were at worst 1.007.
4.A.3. Beam profiles
RBD profile data is presented here for all field sizes mea-
sured (2 9 2 cm2 to 28 9 28 cm2) at depths 1.3 cm and
10.0 cm. The vendor used crossline data to model the TPS.
Data presented from each institution had crossline and in-line
measurements averaged to compare with RBD. The dose
regions (high-dose, penumbra, and low-dose) were based on
the location of the inflection points and width of the penum-
bra from the RBD set. The maximum point difference
between the RBD and measured data in the high dose region
is as follows for various field sizes at each institution for
depth = 1.3 cm. These results are also featured above in
Table III. All high-dose region doses matched RBD high-dose
region doses within 1.6% for both institutions. All maximum
point dose differences were within the penumbra region. In
the RBD Data set for depth = 1.3 cm, the penumbra ranged
from 0.3 cm to 0.7 cm for field sizes 2 9 2 to 28 9 28 cm2.
The measured penumbra regions matched RBD penumbra
regions within 2 mm distance to agreement for all field sizes
at depth = 1.3 cm. Low dose tails were within 0.9% (maxi-
mum point difference at all points) of RBD low dose tails for
both institutions, except for one point (5.3%, UPENN) for a
point on the edge of the penumbra (2 9 2 cm2).
The maximum point difference between the RBD and
measured data in the high dose region is as follows for vari-
ous field sizes (cm2) at each institution for depth = 10.0 cm.
These results are also featured above in Table III. All highest
maximum point dose differences were within the penumbra
region. In the RBD set for depth = 10.0 cm, the penumbra
ranged from 0.4 cm to 0.8 cm for field sizes 2 9 2 to
28 9 28 cm2. The measured penumbra regions matched
RBD penumbra regions within 2 mm distance to agreement
for all field sizes at depth = 10.0 cm. Low dose tails were
within 1.7% (maximum point difference at all points) of
RBD low dose tails for both institutions, except for one point
(4.7%, UPENN) for a point on the edge of the penumbra
(2 9 2 cm2).

Beam profile Beam profile

High dose Region High dose Region 4.B. Output factor measurements
PDD d = 1.3 cm d = 10.0 cm
For all symmetric and asymmetric output factors mea-
FS [cm2] MDACC UPENN MDACC UPENN MDACC UPENN sured, the maximum ratios between the measured data and
the RBD were 1.008 and 1.009 for each institution (MDACC,
2 –0.4 –1 –1.3 –1.6 –1.3 –2.1
UPENN), respectively, and for field sizes >2 9 2 cm2.
4 –0.4 –0.3 –0.5 –0.3 –0.6 –0.5
6 –0.3 0.4 –0.7 –0.1 –0.8 –0.1
8 –0.3 –0.6 0.3 –0.2 –0.3 –0.3
4.C. Mechanical system validation
10 –0.4 –0.4 0.4 –0.3 0.5 –0.1
20 0.3 0.3 0.7 –0.4 0.7 N/A 4.C.1. Lasers (virtual-iso) versus machine treatment
28 –0.3 0.5 1.0 0.4 0.7 N/A isocenter

MDACC, MD Anderson cancer center; PDD, percentage depth dose; RBD, repre- Each of the three lasers indicating the virtual isocenter
sentative beam data set; UPENN, university of pennsylvania. was tested independently. All virtual isocenter laser axes were

Medical Physics, 46 (10), October 2019

4310 Netherton et al.: Commissioning the Halcyon 4310

found to indicate the treatment isocenter position to within

1 mm accuracy.
4.C.2. Gantry and collimator angle indicators
Comparison of the measured gantry angles from the spoke
shot on the EBT3 film, referenced to marked bubble level
line, to the digital readout are within 0.2°.
Gantry angle indicator accuracy was also tested by
observing fiducial movement in portal images under verti-
cal couch motion. On average, gantry angles were veri-
fied to be within 0.19° of expected gantry angles. When
1 mm setup error was introduced, gantry angles were cal-
culated to be within 0.49° of expected angles, on aver-
age, showing that, even in the presence of a small setup
error, this test is reasonable for checking that the gantry
angles are within specifications. For the collimator angle
indicator accuracy testing, the results show that the mea-
sured collimator angles are 0.13° (different from digital
setting by 0.13°) and 89.86° (different from digital setting
by 0.14°).
4.D. 4.3. Imaging system validation and
performance
For high-dose MV-MV mode, the mean extra-target
doses to the heart, left lung, right lung and spine were
1.18, 1.64, 0.80 and 1.11 cGy per image set, respectively.
The mean in-target doses were 3.36, 3.72, 2.61 and
2.69 cGy per image set, respectively. For high-dose MV-
CBCT mode, the mean imaging doses to the heart, left
lung, right lung and spine were 8.45, 7.16, 7.19 and
6.51 cGy per image, respectively. The accuracy of TPS
calculated imaging dose compared to measured dose, on
the anthropomorphic phantom gave average differences of
0.05, 0.35, 0.35, 0.09 and 0.01 cGy for the heart,
left lung, right lung, spine, and breast, respectively. For
more details on our exhaustive study into the imaging
dose with the Halcyon, readers are referred to Li et al.17
and Mihailidis et al.18
4.E. Beam model verification, end-to-end testing,
and treatment plan validation
For beam model validation, the MPPG5a recommended
limit for measured dose distributions matching the TPS is 2%
within the high dose region, 3 mm distance to agreement
within the penumbra region, and 3% in the low-dose tails
region. Detector choices as a function of institution and FS
are specified in Table II. For all field sizes measured
(2 9 2 cm2 to 28 9 28 cm2), PDD curves from both institu-
tions matched the exported TPS data within a maximum
point difference of 1.0% (step size = 1 mm). For field sizes
greater than 2x2 cm2, this maximum point difference was
0.7%. The maximum point difference between the TPS data
and measured PDD data (Table III) is as follows for various
field sizes (cm2) at each institution.
TPS profile data is presented here for all field sizes mea-
sured (2 9 2 cm2 to 28 9 28 cm2) at depths 1.3 cm and
10.0 cm. Data presented from each institution had crossline
and in-line measurements averaged to compare with TPS data.
The dose regions (high-dose, penumbra, and low-dose) were
based on the location of the inflection points and width of the
penumbra from the TPS data set. The maximum point differ-
ence between the TPS data and measured data (Table IV) in
the high dose region is as follows for various field sizes (cm2)
at each institution for depth = 1.3 cm. All high-dose region
doses matched TPS high-dose region doses within 3.3% for
both institutions on a point-by-point basis. All highest maxi-
mum point dose differences were within the penumbra region.
In the TPS data set for depth = 1.3 cm, the penumbra ranged
from 0.3 cm to 0.5 cm for field sizes 2 9 2 to 28 9 28 and
was tighter than the penumbra from the RBD and institutional
measured data. The measured penumbra regions matched
TPS penumbra regions within 1 mm distance to agreement
for all field sizes at depth = 1.3 cm. Low dose tail point dif-
ference from the TPS (Table IV) is as follows for both institu-
tions for depth = 1.3 cm.
The maximum point difference between the TPS data and
measured data (Table IV) in the high dose region is as follows
for various field sizes (cm2) at each institution for

TABLE IV. Maximum percent difference, measured data vs TPS.

Beam profile Beam profile Beam profile Beam profile

High dose region High dose region Low dose tails Low dose tails
PDD d = 1.3 cm d = 10.0 cm d = 1.3 cm d = 10.0 cm
2
FS [cm ] MDACC UPENN MDACC UPENN MDACC UPENN MDACC UPENN MDACC UPENN

2 0.3 –1.0 –1.3 –0.1 –1.1 –3.3 0.0 0.4 –2.7 6.8
4 –0.3 0.4 0.4 –0.3 –1.4 –4.9 –2.9 1.0 –3.8 3.8
6 –0.4 0.3 –1.0 –2.5 –1.0 –4.7 –4.3 4.8 –2.6 4.0
8 0.6 –0.7 –0.8 –0.3 –5.8 –4.2 4.1 –0.5 8.9 3.2
10 0.5 0.4 –2.6 –1.4 –4.1 –4.2 –6.0 –1.8 –9.0 –2.4
20 0.6 0.4 1.1 0.7 –3.3 N/A 3.1 0.5 –2.4 N/A
28 0.6 0.7 –3.3 –1.6 –2.9 N/A 5.4 2.2 6.8 N/A

MDACC, MD Anderson cancer center; PDD, percentage depth dose; TPS, treatment planning system; UPENN, university of pennsylvania.

Medical Physics, 46 (10), October 2019

4311 Netherton et al.: Commissioning the Halcyon 4311

depth = 10.0 cm. When a diode was used, point-dose agree- TABLE VII. Results of the IROC independent evaluation of the delivered dose
ment was within 1.4% in the high dose-region for field sizes profiles (head and neck phantom).
2 9 2 to 6 9 6 cm2 (MDACC). All high-dose region doses
Film plane Gamma indexa Criteria Acceptable
matched TPS high-dose region doses within 5.8% for both
institutions on a point-by-point basis for larger field sizes Axial 99% 85% Yes
using ion chambers (CC13 and CC04). All highest maximum Sagittal 99% 85% Yes
point dose differences were within the penumbra region. In a
Percentage of points meeting Gamma index criteria of 7% and 4 mm.
the TPS data set for depth = 10.0 cm, the penumbra was
0.3 cm for all field sizes. The measured penumbra regions
matched TPS penumbra regions within 1 mm distance to TABLE VIII. Results of the IROC independent evaluation of the delivered
agreement for all field sizes at depth = 10.0 cm. Low dose point doses (thoracic phantom, IMRT plan).
tail point difference from the TPS (Table III) is as follows for
both institutions for depth = 10.0 cm. Location IROC-H vs. Inst. Criteria Acceptable
The results for DLG and MLC transmission for the distal PTV_TLD_sup 0.97 0.92–1.05 Yes
leaf bank are shown in Table V for both institutions along PTV_TLD_inf 0.97 0.92–1.05 Yes
with the values in the pre-configured treatment planning sys-
tem.
The doses delivered to the IROC head and neck phantom
TABLE IX. Results of the IROC independent evaluation of the delivered dose
using a RA plan, as assessed by IROC-Houston, are shown in profiles (thoracic phantom).
Tables VI and VII. The doses delivered to the IROC thoracic
phantom using an IMRT plan, as assessed by IROC-Houston, Film Plane Gamma indexa Criteria Acceptable
are shown in Tables VII and VIII. The phantom irradiation
Axial 98% 80% Yes
results listed in the table above do meet the criteria estab-
Coronal 99% 80% Yes
lished by IROC Houston in collaboration with the coopera-
Sagittal 99% 80% Yes
tive study groups.
Average over 3 planes 99% 85% Yes
Regarding TG-119, confidence limits for cumulative dose
in high and low dose regions were within 3% for IMRT and a
Percentage of points meeting Gamma index criteria of 7% and 5 mm.
RA (Table IX). A negative mean dose (lower measured dose
than calculated) was found for points in the high dose region
for both IMRT and RA (Table X). TABLE X. TG 119 Results.
Patient specific QA results averaged 100% and 99.6% of
IMRT RA
points passing 3%, 3 mm gamma analysis with global nor-
malization for IMRT and RA, respectively. When analyzed High dose Low dose High dose Low dose
using the more stringent 2%, 2 mm, gamma with local nor-
Mean –0.87% 0.08% –0.60% –0.08%
malization, the results averaged 95.9% and 94.4% of points
Standard Deviation 1.08% 0.57% 0.99% 0.82%
Confidence Limit 3.00% 1.19% 2.53% 1.69%
TABLE V. Dynamic leaf gap and MLC transmission measurements.
Mean, standard deviation, and confidence limit of percent difference between cal-
UPENN UPENN culated and measured ion chamber doses relative to prescription dose for IMRT
MDACC Prototype clinical Eclipse and RA TG 119 plans in high and low dose regions.

DLG (mm) 0.08 0.13 0.13 0.10

passing for IMRT and RA, respectively. For RA delivery, the
Transmission (%) 0.5 0.4 0.4 0.5
average passing rate was 99.3 % (96.6–100%) for 3%/3mm
DLG, Dosimetric leaf gap; MDACC, MD Anderson cancer center; UPENN, uni-
global normalization analysis, and 93.1% (83.3–98.9%) for
versity of pennsylvania. 2%/2 mm local normalization. For static gantry IMRT deliv-
ery, the average passing rate was 98.0% (91.3–100%) for 3%/
TABLE VI. Results of the IROC independent evaluation delivered point doses 3 mm, and 77.7% (63.9–92.2%) for 2%/2 mm.
(head and neck phantom, RA plan).

Location IROC-H vs Inst. Criteria Acceptable 5. DISCUSSION

Primary PTV sup/ant 0.99 0.93–1.07 Yes We have described the combined experience of two insti-
Primary PTV inf/ant 0.98 0.93–1.07 Yes tutions in commissioning the Halcyon linac. The Halcyon
Primary PTV sup/post 0.98 0.93–1.07 Yes presents some new challenges in commissioning, mostly
Primary PTV inf/post 1.00 0.93–1.07 Yes related to its new geometry, and the absence of a light field
Secondary PTV sup 0.99 0.93–1.07 Yes and a mechanical distance measuring device. However, new
Secondary PTV inf 0.98 0.93–1.07 Yes methods were developed to use imaging to position the QA
equipment. Once developed, these became very

Medical Physics, 46 (10), October 2019

4312 Netherton et al.: Commissioning the Halcyon
straightforward and easy to use. Other than this, we were able
to use the same equipment to commission the Halcyon as for
other new treatment devices at our institutions.
Although this report is on the commissioning experience
of two institutions, each institution initially implemented its
commissioning process independently. It was only after each
institution had completed their work that the results were
shared and compared. In most cases, both groups used simi-
lar methods and had similar experiences.
The Eclipse system is pre-loaded with a Halcyon beam
model, which reduces the need to collect extensive beam
data. Instead of collecting data to create and test the beam
model in the treatment planning system, the user collects the
relevant beam data to check what is already in the Eclipse
treatment planning system. Because this was our first experi-
ence with this linac, both institutions collected extensive
beam data—although we expect that this could be signifi-
cantly reduced for most institutions. We experienced no
major challenges or disagreements between the measured
data and the data provided by the vendor (RBD and Eclipse
TPS calculated data). A minor difference we noticed, which
was exceeding MPPG5a recommendations, was that maxi-
mum point differences in the high dose region (d = 10 cm)
between measured profiles and exported TPS profiles, were
greater than 2% for multiple field sizes (Table IV). We
believe this to be attributed to the use of ion chambers vs
diodes, as well as to the fact that a misalignment at depth by
1 mm can result in a large point percent difference when
close to gradient regions (in this case, 5% per mm). When
dose regions are analytically defined and plotted, measure-
ments can then be analyzed according to the limits defined in
MPPG5a. In addition, the effect of TPS modeling on the
RBD beam profiles can be visually observed. Penumbra, in
the case of the profiles exhibited in Fig. 1, is effectively
reduced by half (8–3 mm) –perhaps attributed to the removal
of volume dependency from the ionization chamber. In terms
of the difference between detector choice (MDACC vs
UPENN) for scanning measurements, using a CC13 on a
10 9 10 cm2 field (depth = 10 cm) vs a CC04 yielded a
0.2 cm difference in penumbra—small compared to the
dimensions of the field. These differences in penumbra were
small for large field sizes (>8 9 8 cm2), with the larger
chamber having a larger penumbra due to volume averaging.
Likewise, a 0.2 cm difference between MDACC (EDGE
diode) and UPENN (CC04) penumbra was observed for pro-
files at 2 9 2 cm2. This difference in penumbra is significant
due to the fact that it is 10% of the field size. Thus, larger
chambers would not be appropriate to compare dose profiles
calculated in a TPS to those taken with scanner measure-
ments. For small field sizes, diodes more closely matched rel-
ative TPS beam profiles than did a CC04 chamber.
End-to-end tests were all successful, and the results would
have been acceptable for the systems to be released to the
clinic. It is noted that two of the three systems tested in our
work were pre-clinical systems. It is important to note that
while MPPG5a recommends many tests related to static, open
beam characterization, the Halcyon only treats with dynamic
Medical Physics, 46 (10), October 2019
4312
MLC motions. Thus, exhaustive end-to-end tests and initial
characterization of a large range of patient-specific quality
assurance tests may be more reflective of beam model verifi-
cation. Other work that also details clinically acceptable
agreement of validation measurements with the pre-defined
TPS is described by Roover et al and employs end-end tests,
IMRT/VMAT validation, and external dosimetric audits on
the Halcyon system.19 We did not examine the minimum data
requirements for spot-checking this system, and should be the
focus of future work. Since the completion of this work, ongo-
ing quality assurance has not indicated the need to change any
of the pre-determined beam data in the clinical system.
There are some differences between the two pre-clinical
systems and the clinical systems which are now commercially
available, or will become available in the future. One differ-
ence is in the design of the MLCs. The pre-clinical systems
have MLCs which did not have a tongue-and-groove design,
but the clinical systems do have such a design – meaning the
systems that are currently commercially available have a ton-
gue-and-groove design that is similar to those on other linacs
from the same vendor. Transmission measurements from the
pre-clinical machine were 0.5% while clinical systems were
0.4% (Table V). While we did not determine whether this dif-
ference was statistically significant or due to the absence of
tongue-and-groove from the pre-clinical system, transmission
measurements on the clinical Halcyon 1.1 system were within
the range specified by Lim et al in their publication describ-
ing the characterization of the Halcyon MLC system.20 Other
features of the Halcyon can also be expected to change over
time. One key example is the availability of kilo-voltage
imaging. Our test systems, in common with some systems
currently available, only offer megavoltage imaging. The Hal-
cyon is designed, however, to facilitate the addition of kV
imaging with the imaging beamline perpendicular to the
treatment beamline (the same as with most other medical
linacs) – at the time of writing, this has just become commer-
cially available and has been characterized by Cai et al.21
Kilovoltage imaging may allow a means to facilitate equip-
ment setup during commissioning – although we found that
the MV imaging was sufficient, and may be preferable as it
uses the treatment beam line. Results from other commis-
sioning tests are provided in the supplemental material (Data
S1 & Appendix S1).
Another example of changes that can be expected in future
versions of the Halcyon is the capability to treat multi-isocen-
ter plans. This is needed to allow treatment of larger treat-
ment volumes. When that capability is available, it will be
necessary to include that testing in the commissioning pro-
cess, including the end-to-end tests.
6. CONCLUSIONS
We have reported our experiences and results for commis-
sioning the Halcyon unit. Although the user needs to get com-
fortable with the new geometry of this system, we found that
we could use the same equipment to commission the Halcyon
as is used for traditional (C-arm linacs in our clinics). Because
4313 Netherton et al.: Commissioning the Halcyon
penumbra is not defined as is in flattened beams, we recom-
mend analytically calculating beam regions based on inflec-
tion points. Good agreement was found between our
respective experimental data and the treatment planning sys-
tem calculations for the correct use of detector. Thus, we have
shown that the vendor pre-loaded data agree well with the
independent measured ones during the commission process.
This verifies that a data validation instead of a full-data com-
missioning process may be a more efficient approach. This
work should be useful for new users of the Halcyon device.
CONFLICTS OF INTEREST
This work was, in part, funded by Varian Medical Sys-
tems.
a)
Author to whom correspondence should be addressed. Electronic mail:
Mihailidis@pennmedicne.upenn.edu.
REFERENCES
1. Thompson S, Constantin M, Kokkonen P. Eclipse Photon Beam Models
for Halcyon. Varian Medical Systems: Technical Perspectives; 2017.
2. McEwen M, DeWerd L, Ibbott G, et al. Addendum to the AAPM's TG-
51 protocol for clinical reference dosimetry of high-energy photon
beams. Med Phys. 2014;41:041501.
3. Klein EE, Hanley J, Bayouth J, et al. Task group 142 report: quality
assurance of medical acceleratorsa). Med Phys. 2009;36:4197–4212.
4. Almond PR, Biggs PJ, Coursey BM, et al. AAPM's TG-51 protocol for
clinical reference dosimetry of high-energy photon and electron beams.
Med Phys. 1999;26:1847–1870.
5. Smith K, Balter P, Duhon J, et al. Medical physics practice guideline
8.a.: linear accelerator performance tests. J Appl Clin Med Phys.
2017;18:23–39.
6. Smilowitz JB, Das IJ, Feygelman V, et al. Medical physics practice
guideline 5.a.: commissioning and QA of treatment planning dose calcu-
lations - megavoltage photon and electron beams. J Appl Clin Med Phys.
2015;16:14–34.
7. Fontenot JD, Alkhatib H, Garrett JA, et al. Medical physics practice
guideline 2.a: commissioning and quality assurance of X-ray-based
image-guided radiotherapy systems. J Appl Clin Med Phys.
2014;15:4528.
8. Ezzell GA, Burmeister JW, Dogan N, et al. IMRT commissioning: multi-
ple institution planning and dosimetry comparisons, a report from
AAPM task group 119. Med Phys. 2009;36:5359–5373.
9. AGENCY IAE. Commissioning and quality assurance of computerized
planning systems for radiation treatment of cancer. Vienna: International
Atomic Energy Agency; 2004.
10. Das IJ, Cheng CW, Watts RJ, et al. Accelerator beam data commission-
ing equipment and procedures: report of the TG-106 of the therapy phy-
sics committee of the AAPM. Med Phys. 2008;35:4186–4215.
11. Bissonnette JP, Balter PA, Dong L, et al. Quality assurance for image-
guided radiation therapy utilizing CT-based technologies: a report of the
AAPM TG-179. Med Phys. 2012;39:1946–1963.
12. Chang Z, Wu Q, Adamson J, et al. Commissioning and dosimetric char-
acteristics of TrueBeam system: composite data of three TrueBeam
machines. Med Phys. 2012;39:6981–7018.
13. Glide-Hurst C, Bellon M, Foster R, et al. Commissioning of the Varian
TrueBeam linear accelerator: a multi-institutional study. Med Phys.
2013;40:031719.
Medical Physics, 46 (10), October 2019
4313
14. Clivio A, Vanetti E, Rose S, et al. Evaluation of the machine perfor-
mance check application for TrueBeam linac. Radiat Oncol.
2015;10:97.
15. Li Y, Netherton T, Nitsch PL, et al. Independent validation of machine
performance check for the Halcyon and TrueBeam linacs for daily qual-
ity assurance. J Appl Clin Med Phys. 2018;19:375–382.
16. Low DA, Li Z, Drzymala RE. Minimization of target positioning error in
accelerator-based radiosurgery. Med Phys. 1995;22:443–448.
17. Li Y, Netherton T, Nitsch PL, et al. Normal tissue doses from MV
image-guided radiation therapy (IGRT) using orthogonal MV and MV-
CBCT. J Appl Clin Med Phys. 2018;19:52–57.
18. Mihailidis D, Ling CC, Brady L, et al.Evaluation of MV imaging dose
for the first clinical Halcyon system. ESTRO; 2017; Vienna, Austria. p.
PO-0992.
19. De Roover R, et al. Validation and IMRT/VMAT delivery quality of a
preconfigured fast-rotating O-ring linac system. Med Phys.
2019;46:328–339.
20. Lim TY, Dragojevic I, Hoffman D, Flores-Martinez E, Kim GY. Charac-
terization of the Halcyon TM multileaf collimator system. J Appl Clin
Med Phys. 2019;20:106–114.
21. Cai B, et al. Characterization of a prototype rapid kilovoltage x-ray
image guidance system designed for a ring shape radiation therapy unit.
Med. Phys. 2019;46:1355–1370.
SUPPORTING INFORMATION
Additional supporting information may be found online in
the Supporting Information section at the end of the article.
Fig. A1. Las Vegas Phantom images.
Table A1. TG51 measurement for Halcyon.
Table A2. Radiation field size verification for different MLC
settings.
Table A3. Isocenter Verification with Winston-Lutz (radius,
mm).
Table A4. Normalized dose measurements for varying gantry
speed, dose rate and MLC speed.
Table A5. Distance from MV imaging to radiation isocenter
as calculated using the Logos XRV-124 system (Average of
four trials shown).
Table A6. Basic MV imager performance measurements
(MCACC and UPENN).
Table A7. Contrast and spatial resolution measurements (us-
ing CATPHAN).
Table A8. MVCBCT geometric accuracy measurements
(CATPHAN).
Table A9. CT number accuracy measurements using the
CATPHAN (MD Anderson) and QUART phantom
(UPENN).
Table A10. CT image uniformity measurements using the
CATPHAN (MD Anderson) and QUART phantom
(UPENN).
Data S1. Data supplement for PDD, beam profile, output fac-
tors, and surface dose (VARIAN, UPENN, MDACC).
Appendix S1. Supplemental text describing shielding
requirements, output calibration, mechanical tests, radiation
tests, MLC and gantry tests, and imaging system tests
(UPENN, MDACC).
 1 
 

1  APPENDIX

2  A. Shielding requirements.

3  The Halcyon linac has a beam-stop, which should allow for flexibility of installation in an

4  existing vault. Radiation surveys and other measurements should be the same as with other

5  linacs, following applicable State, Federal, and other regulations. A radiation survey was

6  conducted to validate the shielding adequacy. With a lateral beam at maximum field size

7  irradiating a phantom, a maximum exposure rate of ~9 R/hr was measured at 2 m from isocenter

8  in both lateral and longitudinal directions (for a 28x28 cm2 field on a phantom, with the gantry

9  rotated to the contralateral side of the measurement point); the lateral position being downstream

10  of the beam direction. A maximum leakage (all MLC closed) exposure rate of ~2 R/hr was

11  measured at 2 m from the isocenter directly behind the bore and laterally to the gantry with the

12  source proximal to the point of measurement.

13   
14  B. Output Calibration

15  The Halcyon system allows MU calibration to be performed in three alternative geometries. At

16  our institutions we followed the TG51 protocol1 (including the addendum for FFF beams2) to

17  calibrate the Halcyon to 1.00cGy/MU to water, measured in a water phantom, at the depth of

18  maximum dose, with SSD = 100 cm. We used the IBA Blue2 Phantom and Sun Nuclear 1D water

19  phantoms for this experiment. First, the water tank was aligned to the external setup lasers on the

20  treatment table and a water-proof farmer chamber (Exradin A12 / PTW TN30013) was inserted

21  for the output measurements. The water tank was aligned and SSD set as described above. Percent

22  depth dose measurements were performed to determine the beam quality correction factor kQ with

23  and without the lead foil as described.2,1

 2 
 

24  For UPENN Halcyon 1.0, the measurements (Table A1) showed that PDD10=63% and

25  63.9% without and with the lead foil, respectively. Similarly, MDACC Halcyon 1.0 corresponding

26  values differed less than a percent for PDD(10) with and without lead. For Halcyon 1.0, equation

27  (1) and Table I in Ref. 2 were used to calculate kQ for the A12 and TN30013 chambers without

28  lead (0.9974 and 0.9958, respectively) and with lead (0.9962 and 0.9951). This small difference

29  in KQ (with and without lead) is also reported by Lloyd et al3. The correction for change in radial

30  beam profile over the chamber volume (Prp) was 0.25%, calculated by averaging the profile scan

31  results over the chamber length. The output calibration was measured at 10 cm depth and corrected

32  to dmax using the clinical PDD10 for a 10x10 cm2 reference field at 100 SSD, as per TG-51. An

33  adjustment of about 1% was made relative to the output established at installation. For safety

34  considerations, the Halcyon is restricted to a limit of 5% adjustment relative to the factory setting.

35  Varian service must be contacted in that limit needs to be exceeded.

36  C. Mechanical tests

37  Coincidence of radiation isocenter during collimator or gantry rotation

38  Evaluation of the radiation isocenter was performed using the “star-shot” method with EBT3 films.

39  For the gantry rotation star shot, the EBT3 film was positioned vertically in the plane of the

40  isocenter. For the collimation star shot the film was horizontal at source-to-film distance 100cm.

41  An in-house MLC QA software was used for analysis. The tolerance was: ±2 mm (non-

42  SRS/SBRT).4 For collimator rotation, the radiation isocenter for collimator rotation was found

43  within 0.3 mm diameter circle (both institutions). For gantry rotation, the radiation isocenter for

44  gantry rotations was found within 0.8 mm diameter circle (MDACC) and 0.4 mm diameter circle

45  (UPENN).
 3 
 

46  MPC performs isocenter size checks and combines collimator rotations and gantry

47  rotations. We have assessed the ability of the MPC to correctly identify intentionally inserted errors

48  for the Halcyon unit – this data is reported elsewhere5. The average isocenter size (radius, the size

49  of treatment isocenter is defined as the maximum distance of the beam’s central axis from the

50  idealized isocenter) of 30 measurements is 0.79 mm which is twice that found with film. A possible

51  explanation for this disagreement could be due to the sparsity of gantry angles chosen with film.

52  Though the angles chosen were in 30o increments, discrete angles with the largest

53  deviations from isocenter could be missed (see maximum deviations depicted in Section describing

54  Isocenter Verification using WL test). In addition, star shots do not fully capture gantry wobble in

55  the in/out direction. It is suggested that sparse, film-based star shots alone should not be used for

56  this reason. A WL-type test is recommended as described below in the “Isocenter Verification”

57  section.

58 

59  Table travel

60  The measurement of the physical treatment table travel was performed by using a ruler versus

61  the machine lasers of 10.0 cm movements in each direction. The treatment table was moved by

62  digital setting in the system. The actual travel distances were measured using the ruler. The

63  results for the difference of the distance measured by using ruler and digital setting for each

64  movement revealed agreement better than 1 mm for all travel axes.

65 

66  Digital distance indicators.

 4 
 

67  Known and measured distances (using digital tools at the linac console) were compared based on

68  images of a polystyrene board with graticules to assess scaling. The results showed that the

69  differences between measured and expected points 10 cm apart are less than 0.5 mm.

70 

71  D. Radiation tests

72  Radiation field size verification

73  The graticule board was placed perpendicular to the beam central axis at SSD=100 cm, collimator

74  at 0o. Metal BBs were placed on the graticules at known positions. Images of the board with BBs

75  for different field size settings were acquired. The distance from field edge to the center of the BB

76  on images was measured for symmetric fields: 1010 cm2, 2020 cm2 and 2626 cm2.

77  For asymmetric fields, MLCs were digitally set for different field size openings, and

78  images were acquired. The distance from the center crosshair to the field edges was measured

79  manually using digital measurement tools in the service mode. The difference between digital

80  settings and the measured distances were also calculated.

81  We found the agreements were within 1.3 mm for all fields, as seen in Table A2. Small

82  uncertainty in the values on the order of 0.5 mm is expected since all BB’s were hand-placed.

83  Using a graticule board with machined high-contrast inserts would be desirable for more accurate

84  measurements.  

85  Isocenter Verification

86  The WL pointer phantom was aligned to the isocenter. Then, 36 images at different gantry angles

87  (from 0o to 350o, 10o increment) and collimator at 0o were acquired using MV imaging system.

88  The Sun Nuclear Machine software was used to analyze the images. The method used to compute

89  3D target position error is based on the study by Low, et al.6. Y-axis is the gantry rotation axis, Z-
 5 
 

90  axis coincides with the central axis of the radiation beam, pointing into the collimator, and X is

91  assigned to form a right-hand coordinate system. The origin is defined as the center of each

92  imaging field. The horizontal, vertical and total 2D offsets versus gantry angle were determined.

93  Although the WL test result were within the 2 mm tolerance, agreement is better at the cardinal

94  angles than at intermediate angles, especially for X-axis. In general, a sparse, film-based star-shot

95  test alone is not recommended, but if further verification of the WL test are desired, performing a

96  star-shot at the angles with maximum offset is suggested. Isocenter verification the results are

97  given in Table A3.

98 

99  E. MLC and Gantry tests

100  Picket fence test

101  A picket fence pattern was delivered to the EPID using dynamic QA fields to assess MLC

102  alignment and picket position accuracy. The picket fence pattern images were analyzed using in-

103  house software to evaluate the picket spacing and slope relative to the imager coordinates

104  (MDACC). Both proximal and distal banks of leaves were used for “2-bank” tests, while only the

105  distal bank of leaves were used for the “lower-bank” tests. Static tests were performed at the

106  cardinal gantry angles, and a dynamic test was performed under RA delivery. The static MLC

107  picket fence testing results for different gantry angles revealed a maximum picket spacing

108  difference of 0.7 mm, and a maximum slope of 0.17 degrees. The RA MLC picket fence test

109  resulted in a maximum picket spacing difference of 0.4 mm, and a maximum slope of 0.21 degrees.

110 

111  MLC speed, gantry speed and dose rate testing

 6 
 

112  Patterns of equivalent dose delivered under varying levels of dose rate, gantry speed, and MLC

113  speed were captured by the portal imager, and normalized to an open-field image.7 Three test files

114  were provided by Varian for delivery of dose rate levels from 128-800 MU/min, gantry speeds

115  from 3.7-23 deg/s, and MLC speeds from 1-5 cm/s respectively. The images were analyzed to

116  verify that the normalized dose in each delivered pattern was within the 2% specification. The

117  acquired EPID images were analyzed to determine the normalized dose for each delivered pattern,

118  and the results are presented below for dose rate, gantry speed, and MLC speed (Table A4). The

119  tolerances for these tests are ±2%.

120   

121   

122  F. Imaging system tests

123  Coincidence of radiation isocenter and MV imager isocenter

124  Coincidence of the radiation and imaging isocenters are checked as part of the MPC process.

125  We independently evaluated this using the Logos XRV-124 device (Logos Systems

126  International, Scotts Valley CA). The XRV-124 is a conical scintillator that converts an incident

127  beam into two visible spots in the beam direction that can be used to calculate the location of the

128  beam vector in the device’s native coordinate system. A CT-based treatment plan was created in

129  the Eclipse treatment planning system that consisted of eight 2x2 cm square fields at equally-

130  spaced gantry angles over the full 360° gantry rotation. The clinical IGRT process was used to

131  align the device to the MV imaging isocenter prior to delivery of the test treatment plan. The

132  isocenter location was calculated in the XRV software by finding the average location of the

133  mid-points of the shortest line segment between every set of 2 beams in the test treatment plan

134  described above. Radiation isocenter size and distance to the MV imaging isocenter were also

135  calculated. Four trials of IGRT alignment were performed, with the average of calculated values
 7 
 

136  reported. The isocenter size for all four trials with the XRV-124 system was found to be 0.7 mm.

137  The average distance of the isocenter location to the MV imaging isocenter was 0.65 mm. The

138  principal components of the distance between MV and radiation isocenters is shown in Table A5.

139   

140  MV imager performance

141  The Las Vegas (Portal Vision) phantom was imaged using both high-quality and low-dose

142  techniques to give QA baselines (suggested by MPPG2a8). MV images taken with Halcyon 1.0 are

143  shown in Figure A1, and all visible holes were recorded in Table A6 (Varian’s specification: Big

144  right dot in 5th row is visible). Our results showed that image quality meet Varian’s specification

145  at both institutions.

146 

147  MVCBCT performance

148  Image quality tests for the MVCBCT mode were performed with the Catphan phantom, and with

149  the QUART DVT_VN phantom that ships with the clinical system. Scans were performed with

150  a 20 cm field length for the high-quality and low-dose techniques. These tests included low- and

151  high-contrast resolution, geometric accuracy, CT number accuracy, and CT uniformity.

152  Results of the image quality tests from MDACC are shown in Table A7, Table A8, Table A9,

153  and Table A10. The results from UPENN results were similar.

154 

155  REFERENCES 

156  1. Almond PR, Biggs PJ, Coursey BM, et al. AAPM's TG-51 protocol for clinical reference dosimetry
157  of high-energy photon and electron beams. Med Phys 1999;26:1847-70.
158  2. McEwen M, DeWerd L, Ibbott G, et al. Addendum to the AAPM's TG-51 protocol for clinical
159  reference dosimetry of high-energy photon beams. Med Phys 2014;41:041501.
160  3. Lloyd, S. A. M. et al. TG-51 reference dosimetry for the HalcyonTM: A clinical experience. J. Appl.
161  Clin. Med. Phys. 19, 98–102 (2018).
 8 
 

162  4. Klein EE, Hanley J, Bayouth J, et al. Task Group 142 report: Quality assurance of medical
163  acceleratorsa). Medical Physics 2009;36:4197-212.
164  5. Li Y, Netherton T, Nitsch PL, et al. Independent Validation of Machine Performance Check for the
165  Halcyon and TrueBeam Linacs for Daily Quality Assurance. J Appl Clin Med Phys 2018;19:375-382.
166  6. Low DA, Li Z, Drzymala RE. Minimization of target positioning error in accelerator-based
167  radiosurgery. Med Phys 1995;22:443-8.
168  7. Ling CC, Zhang P, Archambault Y, Bocanek J, Tang G, Losasso T. Commissioning and quality
169  assurance of RapidArc radiotherapy delivery system. Int J Radiat Oncol Biol Phys 2008;72:575-81.
170  8. Fontenot JD, Alkhatib H, Garrett JA, et al. AAPM Medical Physics Practice Guideline 2.a:
171  Commissioning and quality assurance of X-ray-based image-guided radiotherapy systems. J Appl
172  Clin Med Phys 2014;15:4528.
173 
174 

175 

176 

177 

178 

View publication stats