Application of the gamma evaluation method in Gamma Knife film dosimetry
Jeong-Hoon ParkDepartment of Neurosurgery, Seoul National University Bundang Hospital, Seongnam 463-707,Korea and Department of Biomedical Engineering, College of Medicine, The Catholic University ofKorea Seoul 137-701, Korea
Jung Ho Han, Chae-Yong Kim, and Chang Wan OhDepartment of Neurosurgery, Seoul National University Bundang Hospital, Seongnam 463-707,Korea and Department of Neurosurgery, College of Medicine, Seoul National University, Seoul 110-799,Korea
Do-Heui LeeDepartment of Neurosurgery, Asan Medical Center, College of Medicine, University of Ulsan,Seoul 138-736, Korea
Tae-Suk Suha)
Department of Biomedical Engineering and Research Institute of Biomedical Engineering,College of Medicine, The Catholic University of Korea, Seoul 137-701, Korea
Dong Gyu Kim and Hyun-Tai Chunga)
Department of Neurosurgery, College of Medicine, Seoul National University, Seoul 110-799, Korea
(Received 15 November 2010; revised 22 June 2011; accepted for publication 28 August 2011;
published 27 September 2011)
Purpose: Gamma Knife (GK) radiosurgery is a minimally invasive surgical technique for the treat-ment of intracranial lesions. To minimize neurological deficits, submillimeter accuracy is required
during treatment delivery. In this paper, the delivery accuracy of GK radiosurgery was assessed
with the gamma evaluation method using planning dose distribution and film measurement data.
Methods: Single 4, 8, and 16 mm and composite shot plans were developed for evaluation usingthe GK Perfexion (PFX) treatment planning system (TPS). The planning dose distributions were
exported as digital image communications in medicine radiation therapy (DICOM RT) files using
a new function of GK TPS. A maximum dose of 8 Gy was prescribed for four test plans. Irradiation
was performed onto a spherical solid water phantom using Gafchromic EBT2 films in the axial and
coronal planes. The exposed films were converted to absolute dose based on a 4th-order polynomial
calibration curve determined using ten calibration films. The film measurement results and planning
dose distributions were registered for further analysis in the same Leksell coordinate using in-house
software. The gamma evaluation method was applied to two dose distributions with varying spatial
tolerance (0.32.0 mm) and dosimetric tolerance (0.32.0%), to verify the accuracy of GK radiosur-
gery. The result of gamma evaluation was assessed using pass rate, dose gamma index histogram
(DGH), and dose pass rate histogram (DPH).
Results: The 20, 50, and 80% isodose lines found in film measurements were in close agreementwith the planning isodose lines, for all dose levels. The comparison of diagonal line profiles across
the axial plane yielded similar results. The gamma evaluation method resulted in high pass rates of
>95% within the 50% isodose line for 0.5 mm=0.5% tolerance criteria, in both the axial and coro-nal planes. They satisfied 1.0 mm=1.0% criteria within the 20% isodose line. Our DGH and DPHalso showed that low isodose lines exhibited inferior gamma indexes and pass rates compared with
higher isodose lines.
Conclusions: The gamma evaluation method was applicable to GK radiosurgery. For all test plans,planning dose distribution and film measurement met the tolerance criteria of 0.5 mm=0.5% withinthe 50% isodose line which are used for marginal dose prescription VC 2011 American Associationof Physicists in Medicine. [DOI: 10.1118/1.3641644]
Key words: radiosurgery, Gamma Knife Perfexion, film dosimetry, gamma evaluation, gamma
index
I. INTRODUCTION
Stereotactic radiosurgery (SRS) is a noninvasive surgical
technique used for the delivery of highly collimated radia-
tion in small fractions. After its initial introduction in the
1950s by Lars Leksell,1 it was widely spread by the develop-
ment of the linear accelerator SRS technique in the 1980s.2
Exclusive SRS modalities, such as Gamma Knife (GK),
(Elekta AB, Stockholm, Sweden), CyberKnife (CK, Accuray
Inc., Sunnyvale, CA), and Novalis (Varian Medical Systems
5778 Med. Phys. 38 (10), October 2011 0094-2405/2011/38(10)/5778/10/$30.00 VC 2011 Am. Assoc. Phys. Med. 5778
Inc., Palo Alto, CA) system, rendered SRS as a major treat-
ment method in both radiation oncology and neurosurgery.
Tomotherapy is also expanding its application to SRS.36
In particular, GK SRS targets intracranial lesions, such as
brain tumors, vascular diseases, and some functional
diseases.710 The presence of many critical structures and el-
oquent areas in the human brain requires careful SRS treat-
ment planning as well as dosimetric verification of the plan.
The verification between the plan and delivery is achieved
by comparing the planning dose distribution exported from
the treatment planning system (TPS) with the measurement
data obtained using radiation detectors. Although this was
studied extensively in radiation therapy physics, only a few
studies are available on GK SRS. These studies used mainly
a polymer gel to obtain a volumetric dose distribution11,14 or
to measure the output factors of collimators.15 Radiochromic
films also began to be accepted as dosimetric tools in GK.
Yamaguchi et al.16 compared absolute linear dose profiles inhead phantom with TPS data for four collimators of GK using
GafchromicTM (International Specialty Products, Wayne, NJ)
MD-55 films and a flatbed scanner. Novotny et al.17 measuredrelative output factors for small collimators of GK Perfexion
via absolute film dosimetry using Gafchromic EDR2, EBT,
and MD-55 films. However, as the direct acquisition of calcu-
lated dose distribution is not allowed in GK TPS,13 more
detailed comparison studies are not available.
In radiation therapy physics with a general linear acceler-
ator, dosimetric comparison of the treatment plan and deliv-
ered dose distribution is achieved using various
sophisticated techniques and tools. Point dose measurement
using an ion chamber, diode detector, or glass rod detector,
etc., is a basic technique for quality assurance (QA),18 this is
also available in GK SRS. Polymer gel dosimetry allows the
calculation of a three-dimensional dose distribution, despite
the necessity of using magnetic resonance imaging (MRI)
scanning for the readout. Recently, two-dimensional array
detectors such as MAPCHECKTM (Sun Nuclear Corp.,
Melbourne, FL) or MATRIXXTM (IBA Dosimetry America,
Bartlett, TN) started to be used commonly for daily and
patient-specific QA purposes, because of the simplicity of
measurement and high accuracy associated with these meth-
ods.19 They support direct comparison of the detector output
with planning dose distribution from TPS or measured dose
distribution from film measurement. Film dosimetry using
radiochromic films, such as Gafchromic MD-55, EBT, and
EBT2 films, provides a better spatial resolution compared
with other tools. In addition, as radiochromic films have a
water-equivalent electron density and two-dimensional dose
distributions are easy to obtain, they are accepted for both
relative and absolute dosimetry.16,20,21 Moreover, many
commercial software programs, such as OMNIPROTM (IBA Do-
simetry America, Bartlett, TN), FILMQATM (International Spe-
cialty Products, Wayne, NJ), and RAYTM (Standard Imaging,
Inc., Middleton, WI), provide a function that allows compre-
hensive analyses of treatment plan, array detector measure-
ments, and film measurements.
In GK SRS, film dosimetry is the only method that can
measure planar dose distribution accurately, as diode or ion
chamber array detectors for GK are not available currently. In
addition, as the planning dose distribution is only attainable
via screen snapshot of TPS, accurate dosimetric studies were
impossible in GK. The direct export of three-dimensional dose
distributions in the digital image communications in medicine
radiation therapy (DICOM RT) format has become available
with the development of the GK TPS LEKSELL GAMMAPLANVR
(LGP) 9.0 in late 2009. However, the commercial software
available does not support postprocessing of GK DICOM RT
data for dosimetric studies. In this study, the delivery accuracy
of GK SRS was evaluated using film measurement data and
planning dose distribution obtained with the new LGP feature.
The comparison results were quantified using our in-house
software and the gamma evaluation method devised by Low etal.22 The gamma evaluation method quantifies the spatial anddosimetric differences between reference and evaluated dose
distributions using a single metric called gamma (c) index. Weanalyzed the characteristics of GK dose distributions compre-
hensively using various tolerance criteria, including fundamen-
tal qualitative comparison results.
II. MATERIALS AND METHODS
II.A. Test treatment plan
Four test treatment plans were generated in LEKSELL GAM-
MAPLAN (LGP) version 9.0 TPS. As the Leksell Gamma
Knife PerfexionVR
(LGK PFX) system has three collimator
sizes of 4, 8, and 16 mm,23,24 single 4, 8, and 16 mm shots
were set at Leksell coordinates (100, 100, 100) (Fig. 1) for
three individual plans. For the evaluation of composite shots,
FIG. 1. Drawings of Leksell coordinates: the superior posterior right corner of
the patient is the origin (0, 0, 0) and the center coordinate of the Leksell frame
is (100, 100, 100). The x axis has a right to left direction, the y axis has a pos-terior to anterior direction, and the z axis has a cranial to caudal direction rela-tive to the patient. [From Elekta AB, Online Reference Manual Leksell
GammaPlanVR
8.3 (Document number 1008129), Rev. 01, Copyright VC 2008/
06 by Elekta Instrument AB. Reprinted by permission of Elekta AB].
5779 Park et al.: Gamma evaluation method in GK dosimetry 5779
Medical Physics, Vol. 38, No. 10, October 2011
which is a characteristic feature of LGK PFX, a single shot
of the [8 B 4 16 8 B 4 16] collimator sector arrangement was
added as the fourth plan (Fig. 2). The prescription dose was
8 Gy at the 100% isodose line for all plans, as the response
of EBT2 film is more sensitive below the 10 Gy exposure
dose.
For all test plans, three-dimensional dose distributions
inside the dose calculation matrix were exported as an RT
dose module of the DICOM RT standard using the DICOM
RT extension of TPS. The size of dose calculation matrixes
was set to minimum, just covering the 10% isodose line of
each shot, to assure a dose calculation that was as accurate as
possible. Therefore, the exported dose volume ranged from
15 15 15 mm for the 4 mm plan to 60 60 60 mmfor the 16 mm plan. The dose calculation grid size was set to
0.1 mm for the 4 and 8 mm plans and to 0.2 mm for the other
plans, to maintain a similar number of reference points inside
isodose lines.
II.B. Gamma Knife film dosimetry
Film measurement was performed on a spherical phantom
with a diameter of 16 cm to obtain delivered dose distribu-
tions. Elekta manufactures two different phantoms for LGK
PFX: a polystyrene phantom (Fig. 3(a)) and a solid water
phantom [Figs. 3(b), 3(c)].25 Both can hold a film and an ion
chamber for dosimetric purposes. However, they are made
of different materials and have different fixation structures
onto the couch. The polystyrene phantom has a structural
problem that attenuates radiation output from the lateral
direction, as pointed out by Bhatnagar et al.26 It also yields aradiation output that is 2% lower because of the higher elec-
tron density of polystyrene compared with water.27 To over-
come these disadvantages, we used a solid water phantom
that was newly designed for LGK PFX. It is made of pure
solid water material and does not block any collimator sec-
tor. It can hold a maximum of ten measurement films and
can be configured in both the axial (xy) and coronal (xz)planes. It provides a film perforator to align the films in the
stack so that they are compressed evenly. Two perforated
holes are separated by 52 mm at (1006 26, 100, 100) per-pendicular to the cranialcaudal axis [Fig. 3(d)]. These were
used for the alignment with the plans after image processing.
To minimize alignment errors between film and calcula-
tions, we performed another test plan using stereotactic CT
images of the spherical phantom. After defining the Leksell
coordinate of phantom images, the exact center of the film
stack was found in TPS by calculating the center of the fixa-
tion bars. There was no discrepancy between the center of
the film stack and the center of coordinates (100, 100, 100),
which allowed the minimization of alignment errors. There-
fore, the film was guaranteed to coincide with the plan and
only systematic errors occurred. The long-term accuracy and
stability of the positioning in GK have been reported by
Heck et al. as being within 0.2 mm.28
We used Gafchromic EBT2 film in all measurements
[Fig. 3(d)]. The EBT2 film has enhanced physical character-
istics compared with the previous EBT film, such as lower
sensitivity to light, lower postexposure, and stronger resist-
ance to physical damage. The detailed explanation and
experiments using the new EBT2 film are found in Richley
et al.29 and Andres et al.30 The films were cut into a size of60 60 mm along the same orientation as that of the originalfilm to be fitted into the phantom. After perforation, films
were placed on three successive stacks in the center of the
phantom, i.e., y 95, 100, and 105 or z 95, 100, and 105.First, calibration films were irradiated to obtain optical den-
sity to absolute dose correlation. Eight films were exposed to
0.5, 1, 2, 3, 4, 6, 8, and 10 Gy using the 16 mm collimator
and one film was left unexposed. The films were irradiated
using the physics mode of LGK PFX for an exposure time
that was determined as the irradiation dose divided by the
current dose rate. The exposure time was determined to be
within 0.01 min. Next, the four treatment plans developed in
Sec. II A were delivered in the axial (xy) and coronal (xz)configurations of the phantom; thus, eight studies were per-
formed. Both calibration and measurement film examina-
tions were repeated using five film sets to average out
random noise caused by the experimental setup.
Forty-eight hours after the exposure, the films were
scanned as TIFF images at a resolution of 300 DPI (0.08
mm=pixel) and a bit color depth of 48 (16 bit=channel) usingthe commercial flatbed scanner EPSON Expression 10 000
XL (Epson America Inc., Long Beach, CA) with a transpar-
ency unit. After a scanner warming time of 30 min, the cali-
bration and measurement films were scanned five times to
minimize the random noises and uncertainties that occurred
during the scanning procedure. For all calibration films, av-
erage pixel values located within a 3 mm radius from the
center were analyzed with the automatic region of interest
(ROI) function of our software. It maintains a consistent
ROI location for reliable analysis using the coordinate infor-
mation determined from the punch holes. They were con-
verted to optical density and calibration curves of the 4th-
order polynomial were determined for each RGB channel
FIG. 2. Collimator sector configuration for the composite shot, which is a
characteristic feature of GK PFX. The 4, 8, and 16 mm shots and blocked
sectors were mixed.
5780 Park et al.: Gamma evaluation method in GK dosimetry 5780
Medical Physics, Vol. 38, No. 10, October 2011
using the curve fitting function of IMAGEJ (National Institutes
of Health, Bethesda, MD) software. For further analysis, the
measurement films were converted to absolute dose using
the calibration curve. A detailed film dosimetry procedure is
described in Devic et al.,20 we adopted similar procedures.
II.C. Application of the gamma evaluation method
First, the plan and measurement data should be aligned in
the same Leksell coordinate, as they have different centers,
sizes, and rotation angles. We developed an in-house soft-
ware using MATLAB R2010 (The Mathworks, Natick, MA) to
align and scale them into the same coordinates. The exact
center of the measurement film was equivalent to the center
of two punching holes used for fixation, which can be deter-
mined by detecting the perimeter of the circle. In addition,
the rotation angle of the film during the scanning procedure
can be deduced from the line connecting the two centers.
The film images were transformed to Leksell coordinates
using the coordinate center, rotation angle, and known image
resolution. As the center voxel of RT dose data in the
DICOM RT file already corresponded to the center of the
Leksell coordinate in the planning stage, we now had two
dose distributions in the same Leksell coordinate, which
were used for further comparison. Before comparing two
data sets, we applied a Wiener filter with a mask size of
7 7 pixels to measurement data, to minimize image noisescaused by the film itself, while preserving the original
data.20 The punch hole regions were excluded in the analysis
because the cutting edge of these holes has an abnormally
high optical density and the hole itself exhibits an all white
background.
As explained in Sec. I, the gamma evaluation method is a
technique used to quantify the similarity of two dose
distributions as a gamma index. It is a composite analysis
that combines both dosimetric and spatial accuracy. It pro-
vides a reasonable measurement of the quantitative compari-
son of reference and evaluated dose distributions. Many
modified gamma evaluation methods have been developed
to accelerate or improve the original algorithm.31,36
Generally, gamma evaluation is achieved by comparing
three-dimensional planning dose distributions from TPS
with two- or three-dimensional dose distributions obtained
using other plans, film measurement, or an array detector.
According to the definition of the original gamma evaluation
method, the gamma index c at the point rr of the referencesystem (rr, Dr) is calculated from Eqs. (1) and (2) for theevaluation system (re, De).
32
C~re; r!r
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffir!e
r!r2Dd2
De r!
e Dr r!2
Dd2
s(1)
c r!r minfC r!e; r!rg8f r!eg: (2)where Dd and DD are the spatial and dosimetric tolerancecriteria for the comparison systems typically set as 3
mm=3%. In gamma evaluation, if c 1, the reference pointrr satisfies the tolerance criteria relative to the evaluationsystem; otherwise, it fails. If the pass rate, which is a ratio of
points that satisfy the tolerance criteria, is higher than 95%,
the two comparison systems are said to be equivalent, at
least under the criteria. Here, we selected DICOM RT data
as the reference data and film measurement as the evaluation
data.
FIG. 3. Dosimetric phantoms used for GK QA purpose
and irradiation sample of the EBT2 film. (a) Old poly-
styrene phantom and phantom holder; (b) new solid
water phantom and EBT2 film in the axial plane config-
uration; (c) in the coronal plane configuration; and (d)
irradiation sample of Gafchromic EBT2 film.
5781 Park et al.: Gamma evaluation method in GK dosimetry 5781
Medical Physics, Vol. 38, No. 10, October 2011
We applied four criteria of 2.0 mm=2.0%, 1.0 mm=1.0%,0.5 mm=0.5%, and 0.3 mm=0.3% to verify the delivery accu-racy of GK SRS. We used Lows original algorithm to
implement the gamma evaluation method. The evaluation
points within a 5 mm distance from the reference point were
included in the evaluation to reduce calculation time. In the
GK plan, dose prescription is given on a much lower isodose
line, e.g., 50%, compared with the radiation therapy plan,
which uses the isodose line of >95%.37 This means that theGK plan has a more inhomogeneous dose distribution inside
the PTV, ranging from 50 to 100%. Therefore, we calculated
the pass rate of reference points inside the 50% isodose line,
which is clinically important, as well as the 20% isodose
line, which was selected for peripheral dose assessment.
However, as other higher and lower isodose lines are com-
monly used to examine hot spots and normal tissue damage,
a single gamma index inside the 20 and 50% isodose lines
cannot explain heterogeneous delivery characteristics. This
led to the design of the dose gamma index histogram (DGH)
and dose pass rate histogram (DPH), which show the relative
delivery accuracy of the plan. First, isodose levels of
10100% were binned into a 5% interval and reference
points inside each bin were collected. Next, the gamma
indexes and pass rates were calculated separately for each
bin and the results were displayed as dose vs pass rate rela-
tion. The DGH and DPH allow the determination of which
isodose range had relatively good delivery accuracy or not.
Generally, gamma evaluation requires reference data with
smaller, i.e., less data points, or equal dimensions compared
with the evaluation data, to prevent interpolation artifacts.32
Our data satisfied this requirement, as DICOM RT dose data
had a resolution of 0.100.20 mm and film measurement had
a resolution of 300 DPI, which is equivalent to 0.08 mm.Moreover, as the pixel spacing of 0.08 mm in the evaluation
data was smaller than the minimum Dd of 0.3 mm, the datasets had a configuration that was suitable for the prevention
of interpolation artifacts.32
FIG. 4. Optical density vs absolute dose curves of EBT2 film in the 010 Gy dose range for red, green, and blue channels.
FIG. 5. Comparison of isodose lines between planning dose distribution and
film measurement in the axial plane. Solid lines are the 20, 50, and 80% iso-
dose lines from the DICOM RT file. Dash lines are measured isodose lines.
5782 Park et al.: Gamma evaluation method in GK dosimetry 5782
Medical Physics, Vol. 38, No. 10, October 2011
III. RESULTS
III.A. Film dosimetry and qualitative comparison
The optical density of calibration films was calculated
from the pixel values of each red, green, and blue channel.
Their correlation to the absolute dose of the 010 Gy range
was determined as a 4th-order polynomial, as shown in Fig.
4. The red and green channels had an R2 value of 0.9999 in
the 010 Gy range; however, it is well known that the red
channel has a good response under a 10 Gy dose in using
EBT2 film.30 Therefore, we used pixel data of the red chan-
nel for further analyses of measurement films. Figure 4
shows that the correlation curve of the red channel was most
sensitive regarding irradiation dose, compared with other
channels in this range.
After alignment of measurement data and DICOM RT
data into the same Leksell coordinate, isodose lines and
linear dose profiles were compared using a qualitative analy-
sis. Figure 5 shows the planning and measured isodose lines
overlaid on the film measurement image. It shows the 20,
50, and 80% isodose lines in axial (xy) plane at z 100. Forall four plans, the 50 and 80% isodose lines exhibited close
agreement between treatment plan and measurement data.
Measurement data exhibited a slightly broader isodose distri-
bution at the 20% isodose line. The disagreement at low iso-
dose lines tended to worsen as the size of the collimator
increased. In the composite shot, isodose lines had elongated
shapes, with variation of major axes depending on isodose
levels. The dose distribution of the composite shot matched
the measurement closely, even in the blocked sector and in
the 16 mm collimator sector direction. Figure 6 compares
the line profiles of treatment plan and measurement data
along the diagonal direction, i.e., from the anterior-right to
the posterior-left side of the patient in the axial plane.
FIG. 6. Comparison of the line profiles of four test plans and film measurements in the diagonal direction along the anterior-right to posterior-left side of the
patient (top left to right bottom side in Fig. 5). The solid lines correspond to DICOM RT data and the dotted lines correspond to film measurement data.
TABLE I. Pass rates of gamma evaluation within the 20 and 50% isodose lines using four evaluation criteria (2.0 mm=2.0%, 1.0 mm=1.0%, 0.5 mm=0.5%, and
0.3 mm=0.3%; units, %).
20% isodose line 50% isodose line
Plan
2.0 mm
=2.0%
1.0 mm
=1.0%
0.5 mm
=0.5%
0.3 mm
=0.3%
2.0 mm
=2.0%
1.0 mm
=1.0%
0.5 mm
=0.5%
0.3 mm
=0.3%
4 mm axial 100 100 94 60 100 100 96 72
coronal 100 100 90 52 100 100 96 68
8 mm axial 100 100 96 67 100 100 98 78
coronal 100 100 95 67 100 100 95 74
16 mm axial 100 100 95 64 100 100 95 76
coronal 100 100 78 49 100 100 97 68
Composite axial 100 100 98 72 100 100 98 80
coronal 100 100 84 50 100 100 98 70
5783 Park et al.: Gamma evaluation method in GK dosimetry 5783
Medical Physics, Vol. 38, No. 10, October 2011
Similar to what was found in the isodose line comparison,
these values agreed closely with regard to the whole isodose
ranges, without distinct mismatched regions.
III.B. Quantitative comparison using the gammaevaluation method
Table 1 summarizes the pass rates of gamma evaluation
using four evaluation criteria for the eight test studies. Inside
the 50% isodose line, all test plans passed the criteria up to
0.5 mm=0.5% in both axial and coronal planes, with pass rates>95%. Axial studies showed higher pass rates of 9498% for0.5 mm=0.5% criteria even within the 20% isodose line. How-ever, other coronal studies satisfied maximum 1.0 mm=1.0%criteria for the 20% isodose line, rather than 0.5 mm=0.5%criteria. Both the 20 and 50% isodose lines failed to achieve
the last criteria of 0.3 mm=0.3%. Figure 7 shows the gammaindex map for 0.5 mm=0.5% tolerance criteria in axial plane.Four plans showed low gamma index values in the irradiated
areas around the center of the shot. In contrast, high gamma
index values (>2.5) were observed in the unexposed areas.This phenomenon was apparent at the borders of 4 and 8 mm
collimator shots and at the blocked sector side of the compos-
ite shot. In the 4 and 8 mm collimator study, DICOM RT data
with a 0.1 mm dose grid were used for the calculation of the
gamma index and a 0.2 mm grid was used for the visualiza-
tion of a coarse gamma index map. The dose grid of 0.1 mm
yielded a pass rate that was 0.5% higher compared with
that obtained using the 0.2 mm grid.
Figure 8 shows the dose gamma index histogram (DGH)
between the 10105% dose ranges, grouped by 5% isodose
bins. For the 0.5 mm=0.5% criteria, the gamma index waskept under 0.6 in most dose bins. High dose bins had a lower
gamma index compared with low dose bins but the differen-
ces were not large. For the 0.3 mm=0.3% criteria, all plansshowed more irregular shapes, with maximum values
exceeding 1.0. Low isodose bins had a relatively higher
gamma index. Figure 9 illustrates the dose DPH representing
individual pass rates within the bin. Similar to what was
observed for the DGH, the DPH revealed the presence of a
nearly flat pass rate distribution for 0.5 mm=0.5% criteria.The DPH of 0.3 mm=0.3% criteria resulted in a low pass ratethat was
conformed to the lesion margin, with the exception of
extremely small lesions. The delivery accuracy inside this
line affects tumor control directly and is associated with nor-
mal brain complications. The results of the analyses
described above show that the planning dose distribution
based on the TPS was in close agreement with the delivered
dose distribution, at least within clinically important ranges.
Results also showed that low isodose levels had lower pass
rates for 0.5 mm=0.5% and 0.3 mm=0.3% criteria comparedwith higher isodose levels, which was also demonstrated
clearly in our DGH and DPH analyses (Figs. 8 and 9). This
was supposedly caused by the sensitivity characteristics of
the EBT2 film; as this film exhibits a relatively steep
response to irradiation doses in the low dose range, as shown
in Fig. 4, small perturbations in absorbed dose can induce
larger differences in film response in the low isodose range
than in the high isodose range. Alternatively, this may be
caused by the inaccuracy of dose models in the peripheral
region, which should be addressed in a future study.
The advances in the research of GK dosimetry have been
relatively poor compared with those of other radiation ther-
apy modalities. In the early days, polymer gels were used
widely as a dosimetric tool, and films are beginning to be
adopted for various measurement purposes. However, as GK
plan data are accessible only via the digitization of the LGP
planning screen or special support provided by the manufac-
turer, research in this area was limited to a few themes. The
new DICOM RT export function is expected to resolve this
problem and motivate researchers. Although official analyti-
cal tools for GK are not available yet, our experiences can be
referenced in the development of customized software. The
Gafchromic EBT2 film and solid water phantom for the GK
also provided a convenient method of measurement of deliv-
ered dose distribution. We expect that this combination can
be extended as a useful research and QA tool for the compar-
ison of dose distributions in GK radiosurgery. In particular,
it is very desirable to include gamma evaluation testing for
certain tolerance criteria in the annual QA procedure of GK.
FIG. 8. Dose gamma index histogram in 5% isodose bins for 0.5 mm=0.5% and 0.3 mm=0.3% tolerance criteria in the axial plane.
Medical Physics, Vol. 38, No. 10, October 2011
5785 Park et al.: Gamma evaluation method in GK dosimetry 5785
In addition, the built-in film analysis feature needs to be sup-
ported in the LGP TPS for routine and convenient analysis.
V. CONCLUSIONS
A comprehensive comparison of planning and delivered
dose distributions in GK SRS was performed using DICOM
RT data, a solid water phantom, and EBT2 film dosimetry.
For single 4, 8, and 16 mm and composite collimator shots,
the comparison of qualitative isodose lines and line profiles
did not reveal any notable deviations between planned and
film measurements. In addition, the gamma evaluation
method was applied to the verification of the quantitative
delivery of GK SRS. It revealed that GK SRS satisfied toler-
ance criteria of 0.5 mm=0.5% inside the 50% isodose line,which is used for marginal dose prescription, and 1.0
mm=1.0% criteria inside the 20% isodose line. Our DPH andDGH analysis also confirmed that low isodose areas
(1040%) exhibited a relatively inferior accuracy compared
with high isodose areas (50100%).
ACKNOWLEDGMENTS
This work was supported by a grant no. 04-2011-
0320110130 from the Seoul National University Hospital
Research Fund and a National Research Foundation of Korea
(NRFK) grant funded by the Korean government (MEST)
(No. 20110001851). The authors acknowledge Mr. Wook
Choi and Mr. Soo-Hyun Lee of Elekta Korea Inc. for their
support on the DICOM RT export procedure. The authors
have no conflict of interest.
a)Author to whom correspondence should be addressed. Electronic
mail: [email protected]; [email protected]. Leksell, The stereotaxic method and radiosurgery of the brain, Acta
Chir. Scand. 102, 316319 (1951).2S. H. Benedict, F. J. Bova, B. Clark, S. J. Goetsch, W. H. Hinson, D. D.
Leavitt, D. J. Schlesinger, and K. M. Yenice, Anniversary Paper: The
role of medical physicists in developing stereotactic radiosurgery, Med.
Phys. 35, 42624277 (2008).3C. Lindquist, Gamma knife radiosurgery, Semin. Radiat. Oncol. 5,197202 (1995).
FIG. 9. Dose pass rate histogram in 5% isodose bins for 0.5 mm=0.5% and 0.3 mm=0.3% tolerance criteria in the axial plane.
Medical Physics, Vol. 38, No. 10, October 2011
5786 Park et al.: Gamma evaluation method in GK dosimetry 5786
http://dx.doi.org/10.1118/1.2969268http://dx.doi.org/10.1118/1.2969268http://dx.doi.org/10.1016/S1053-4296(05)80017-7
4J. R. AdlerJr., S. D. Chang, M. J. Murphy, J. Doty, P. Geis, and S. L. Han-
cock, The Cyberknife: A frameless robotic system for radiosurgery,
Stereotact. Funct. Neurosurg. 69, 124128 (1997).5F. F. Yin, S. Ryu, M. Ajlouni, J. Zhu, H. Yan, H. Guan, K. Faber, J. Rock,
M. Abdalhak, L. Rogers, M. Rosenblum, and J. H. Kim, A technique of
intensity-modulated radiosurgery (IMRS) for spinal tumors, Med. Phys.
29, 28152822 (2002).6E. T. Soisson, N. Hardcastle, and W. A. Tome, Quality assurance of an
image guided intracranial stereotactic positioning system for radiosur-
gery treatment with helical tomotherapy, J. Neuro-oncol. 98, 277285(2010).
7W. Y. Chung, K. D. Liu, C. Y. Shiau, H. M. Wu, L. W. Wang, W. Y. Guo,
D. M. Ho, and D. H. Pan, Gamma knife surgery for vestibular schwannoma:
10-year experience of 195 cases, J. Neurosurg. 102, Suppl: 8796 (2005).8J. H. Han, D. G. Kim, H. T. Chung, C. K. Park, S. H. Paek, C. Y. Kim,
and H. W. Jung, Gamma knife radiosurgery for skull base meningiomas:
Long-term radiologic and clinical outcome, Int. J. Radiat. Oncol., Biol.,
Phys. 72, 13241332 (2008).9J. H. Han, D. G. Kim, H. T. Chung, C. K. Park, S. H. Paek, J. E. Kim,
H. W. Jung, and D. H. Han, Clinical and neuroimaging outcome of cere-
bral arteriovenous malformations after Gamma Knife surgery: Analysis of
the radiation injury rate depending on the arteriovenous malformation vol-
ume, J. Neurosurg. 109, 191198 (2008).10J. H. Han, D. G. Kim, H. T. Chung, S. H. Paek, Y. H. Kim, C. Y. Kim,
J. W. Kim, and S. S. Jeong, Long-term outcome of gamma knife radio-
surgery for treatment of typical trigeminal neuralgia, Int. J. Radiat.
Oncol., Biol., Phys. 75, 822827 (2009).11J. Novotny, Jr., J. Novotny, V. Spevacek, P. Dvorak, T. Cechak, R. Liscak,
G. Brozek, J. Tintera, and J. Vymazal, Application of polymer gel dosim-
etry in gamma knife radiosurgery, J. Neurosurg. 97, 556562 (2002).12P. Karaiskos, L. Petrokokkinos, E. Tatsis, A. Angelopoulos, P. Baras, M.
Kozicki, P. Papagiannis, J. M. Rosiak, L. Sakelliou, P. Sandilos, and L.
Vlachos, Dose verification of single shot gamma knife applications using
VIPAR polymer gel and MRI, Phys. Med. Biol. 50, 12351250 (2005).13P. Papagiannis, P. Karaiskos, M. Kozicki, J. M. Rosiak, L. Sakelliou, P.
Sandilos, I. Seimenis, and M. Torrens, Three-dimensional dose verifica-
tion of the clinical application of gamma knife stereotactic radiosurgery
using polymer gel and MRI, Phys. Med. Biol. 50, 19791990 (2005).14Y. Watanabe, T. Akimitsu, Y. Hirokawa, R. B. Mooij, and G. M. Perera,
Evaluation of dose delivery accuracy of Gamma Knife by polymer gel
dosimetry, J. Appl. Clin. Med. Phys. 6, 133142 (2005).15A. Moutsatsos, L. Petrokokkinos, P. Karaiskos, P. Papagiannis,
E. Georgiou, K. Dardoufas, P. Sandilos, M. Torrens, E. Pantelis, I. Kante-
miris, L. Sakelliou, and I. Seimenis, Gamma knife output factor measure-
ments using VIP polymer gel dosimetry, Med. Phys. 36, 42774287(2009).
16M. Yamaguchi, T. Tominaga, O. Nakamura, R. Ueda, and M. Hoshi,
GAFChromic film dosimetry with a flatbed color scanner for Leksell
Gamma Knife therapy, Med. Phys. 31, 12431248 (2004).17J. Novotny, Jr., J. P. Bhatnagar, M. A. Quader, G. Bednarz, L. D. Luns-
ford, and M. S. Huq, Measurement of relative output factors for the 8 and
4 mm collimators of Leksell Gamma Knife Perfexion by film dosimetry,
Med. Phys. 36, 17681774 (2009).18J. E. Rah, D. O. Shin, J. S. Jang, M. C. Kim, S. C. Yoon, and T. S. Suh,
Application of a glass rod detector for the output factor measurement in
the CyberKnife, Appl. Radiat. Isot. 66, 19801985 (2008).19J. Lee, J. Chung, D. Lee, J. Park, B. Choe, T. Suh, H. Jang, S. Hong, B.
Park, and M. Kang, Discrepancy of IMRT dose delivery due to the dose-
dynamic MLC gravity effect, J. Kor. Phys. Soc. 53, 34363443 (2008).
20S. Devic, J. Seuntjens, E. Sham, E. B. Podgorsak, C. R. Schmidtlein, A. S.
Kirov, and C. G. Soares, Precise radiochromic film dosimetry using a
flat-bed document scanner, Med. Phys. 32, 22452253 (2005).21E. E. Wilcox and G. M. Daskalov, Evaluation of Gafchromic EBT film
for Cyberknife dosimetry, Med. Phys. 34, 19671974 (2007).22D. A. Low, W. B. Harms, S. Mutic, and J. A. Purdy, A technique for the quan-
titative evaluation of dose distributions, Med. Phys. 25, 656661 (1998).23C. Lindquist and I. Paddick, The Leksell Gamma Knife perfexion and
comparisons with its predecessors, Neurosurgery 61, 130140; discussion140131 (2007).
24J. Novotny, J. P. Bhatnagar, A. Niranjan, M. A. Quader, M. S. Huq, G.
Bednarz, J. C. Flickinger, D. Kondziolka, and L. D. Lunsford, Dosimetric
comparison of the Leksell Gamma Knife perfexion and 4C, J. Neurosurg.
109, Suppl: 814 (2008).25J. Novotny, Jr., J. P. Bhatnagar, H. T. Chung, J. Johansson, G. Bednarz, L.
Ma, and M. S. Huq, Assessment of variation in Elekta plastic spherical-
calibration phantom and its impact on the Leksell Gamma Knife calibra-
tion, Med. Phys. 37, 50665071 (2010).26J. P. Bhatnagar, J. Novotny, Jr., M. A. Quader, G. Bednarz, and M. S.
Huq, Unintended attenuation in the Leksell Gamma Knife perfexion
calibration-phantom adaptor and its effect on dose calibration, Med.
Phys. 36, 12081211 (2009).27H. T. Chung, Y. Park, S. Hyun, Y. Choi, G. H. Kim, D. G. Kim, and K. J.
Chun, Determination of the absorbed dose rate to water for the 18-mm
Helmet of a Gamma Knife, Int. J. Radiat. Oncol., Biol., Phys. 79,15801587 (2011).
28B. Heck, A. Jess-Hempen, H. J. Kreiner, H. Schopgens, and A. Mack,
Accuracy and stability of positioning in radiosurgery: long-term results
of the Gamma Knife system, Med. Phys. 34, 14871495 (2007).29L. Richley, A. C. John, H. Coomber, and S. Fletcher, Evaluation and opti-
mization of the new EBT2 radiochromic film dosimetry system for patient
dose verification in radiotherapy, Phys. Med. Biol. 55, 26012617 (2010).30C. Andres, A. del Castillo, R. Tortosa, D. Alonso, and R. Barquero, A
comprehensive study of the Gafchromic EBT2 radiochromic film. A com-
parison with EBT, Med. Phys. 37, 62716278 (2010).31T. Depuydt, A. Van Esch, and D. P. Huyskens, A quantitative evaluation
of IMRT dose distributions: Refinement and clinical assessment of the
gamma evaluation, Radiother. Oncol. 62, 309319 (2002).32D. A. Low and J. F. Dempsey, Evaluation of the gamma dose distribution
comparison method, Med. Phys. 30, 24552464 (2003).33N. J. Lomax and S. G. Scheib, Quantifying the degree of conformity in
radiosurgery treatment planning, Int. J. Radiat. Oncol., Biol., Phys. 55,14091419 (2003).
34S. B. Jiang, G. C. Sharp, T. Neicu, R. I. Berbeco, S. Flampouri, and T.
Bortfeld, On dose distribution comparison, Phys. Med. Biol. 51,759776 (2006).
35M. Wendling, L. J. Zijp, L. N. McDermott, E. J. Smit, J. J. Sonke, B. J.
Mijnheer, and M. van Herk, A fast algorithm for gamma evaluation in
3D, Med. Phys. 34, 16471654 (2007).36M. Chen, W. Lu, Q. Chen, K. Ruchala, and G. Olivera, Efficient gamma
index calculation using fast Euclidean distance transform, Phys. Med.
Biol. 54, 20372047 (2009).37L. Ma, P. Xia, L. J. Verhey, and A. L. Boyer, A dosimetric comparison
of fan-beam intensity modulated radiotherapy with Gamma Knife stereo-
tactic radiosurgery for treating intermediate intracranial lesions, Int. J.
Radiat. Oncol., Biol., Phys. 45, 13251330 (1999).38L. Ma, C. Chuang, M. Descovich, P. Petti, V. Smith, and L. Verhey,
Whole-procedure clinical accuracy of gamma knife treatments of large
lesions, Med. Phys. 35, 51105114 (2008).
Medical Physics, Vol. 38, No. 10, October 2011
5787 Park et al.: Gamma evaluation method in GK dosimetry 5787
http://dx.doi.org/10.1159/000099863http://dx.doi.org/10.1118/1.1521722http://dx.doi.org/10.1007/s11060-010-0227-4http://dx.doi.org/10.3171/jns.2005.102.s_supplement.0087http://dx.doi.org/10.1016/j.ijrobp.2008.03.028http://dx.doi.org/10.1016/j.ijrobp.2008.03.028http://dx.doi.org/10.3171/JNS/2008/109/8/0191http://dx.doi.org/10.1016/j.ijrobp.2009.05.040http://dx.doi.org/10.1016/j.ijrobp.2009.05.040http://dx.doi.org/10.1088/0031-9155/50/6/013http://dx.doi.org/10.1088/0031-9155/50/9/004http://dx.doi.org/10.1120/jacmp.2025.25343http://dx.doi.org/10.1118/1.3183500http://dx.doi.org/10.1118/1.1712393http://dx.doi.org/10.1118/1.3113904http://dx.doi.org/10.1016/j.apradiso.2008.06.041http://dx.doi.org/10.3938/jkps.53.3436http://dx.doi.org/10.1118/1.1929253http://dx.doi.org/10.1118/1.2734384http://dx.doi.org/10.1118/1.598248http://dx.doi.org/10.1227/01.neu.0000289726.35330.8ahttp://dx.doi.org/10.1118/1.3481508http://dx.doi.org/10.1118/1.3093240http://dx.doi.org/10.1118/1.3093240http://dx.doi.org/10.1016/j.ijrobp.2010.05.039http://dx.doi.org/10.1118/1.2710949http://dx.doi.org/10.1088/0031-9155/55/9/012http://dx.doi.org/10.1118/1.3512792http://dx.doi.org/10.1016/S0167-8140(01)00497-2http://dx.doi.org/10.1118/1.1598711http://dx.doi.org/10.1016/S0360-3016(02)04599-6http://dx.doi.org/10.1088/0031-9155/51/4/001http://dx.doi.org/10.1118/1.2721657http://dx.doi.org/10.1088/0031-9155/54/7/012http://dx.doi.org/10.1088/0031-9155/54/7/012http://dx.doi.org/10.1016/S0360-3016(99)00340-5http://dx.doi.org/10.1016/S0360-3016(99)00340-5http://dx.doi.org/10.1118/1.2987669
s1s2s2AF1s2BF2s2CE1E2F3F4F5s3s3AF6T1s3Bs4F7F8s5cor1B1B2B3F9B4B5B6B7B8B9B10B11B12B13B14B15B16B17B18B19B20B21B22B23B24B25B26B27B28B29B30B31B32B33B34B35B36B37B38