Physics in Medicine & Biology

PAPER Related content

- Online pretreatment verification of high-
A novel system for commissioning brachytherapy dose rate brachytherapy using an imaging
panel
applicators: example of a ring applicator Gabriel P Fonseca, Mark Podesta, Murillo
Bellezzo et al.

- Direct reconstruction and associated

To cite this article: Gabriel P Fonseca et al 2017 Phys. Med. Biol. 62 8360 uncertainties of 192Ir source dwell
positions in ring applicators using
gafchromic film in the treatment planning
of HDR brachytherapy cervix patients
O A Awunor, B Dixon and C Walker
View the article online for updates and enhancements. - Performance assessment of the BEBIG
MultiSource high dose rate brachytherapy
treatment unit
Antony Palmer and Bongile Mzenda

This content was downloaded from IP address 80.82.77.83 on 15/11/2017 at 10:56

 Institute of Physics and Engineering in Medicine Physics in Medicine & Biology

Phys. Med. Biol. 62 (2017) 8360–8375 https://doi.org/10.1088/1361-6560/aa8d0a

A novel system for commissioning

brachytherapy applicators: example of a
ring applicator
Gabriel P Fonseca, Michiel R Van den Bosch,
Robert Voncken, Mark Podesta and Frank Verhaegen
Department of Radiation Oncology (MAASTRO), GROW—School for Oncology
and Developmental Biology, Maastricht University Medical Center, Dr. Tanslaan 12,
Maastricht 6229 ET, Netherlands

E-mail: frank.verhaegen@maastro.nl

Received 25 April 2017, revised 25 August 2017

Accepted for publication 15 September 2017
Published 17 October 2017

Abstract
A novel system was developed to improve commissioning and quality
assurance of brachytherapy applicators used in high dose rate (HDR). It
employs an imaging panel to create reference images and to measure dwell
times and dwell positions. As an example: two ring applicators of the same
model were evaluated. An applicator was placed on the surface of an imaging
panel and a HDR 192Ir source was positioned in an imaging channel above
the panel to generate an image of the applicator, using the gamma photons
of the brachytherapy source. The applicator projection image was overlaid
with the images acquired by capturing the gamma photons emitted by the
source dwelling inside the applicator. We verified 0.1, 0.2, 0.5 and 1.0 cm
interdwell distances for different offsets, applicator inclinations and transfer
tube curvatures. The data analysis was performed using in-house developed
software capable of processing the data in real time, defining catheters and
creating movies recording the irradiation procedure. One applicator showed
up to 0.3 cm difference from the expected position for a specific dwell position.
The problem appeared intermittently. The standard deviations of the remaining
dwell positions (40 measurements) were less than 0.05 cm. The second ring
applicator had a similar reproducibility with absolute coordinate differences
from expected values ranging from  −0.10 up to 0.18 cm. The curvature of the
transfer tube can lead to differences larger than 0.1 cm whilst the inclination
of the applicator showed a negligible effect. The proposed method allows the
verification of all steps of the irradiation, providing accurate information about
dwell positions and dwell times. It allows the verification of small interdwell
positions (⩽0.1 cm) and reduces measurement time. In addition, no additional

1361-6560/17/218360+16$33.00  © 2017 Institute of Physics and Engineering in Medicine  Printed in the UK 8360
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

radiation source is necessary since the HDR 192Ir source is used to generate an
image of the applicator.

Keywords: brachytherapy, imaging panel, time-resolve commissioning

S Supplementary material for this article is available online

(Some figures may appear in colour only in the online journal)

1. Introduction

The commissioning of a brachytherapy applicator prior to clinical use is essential to verify

conformance with manufacturer’s specifications assuring the correct treatment delivery and
patient safety. The process may include mechanical inspection, verification of the dwell posi-
tions and dose measurements (Granero et al 2016). The measurements, or at least a sub-set
of them, should be repeated as part of the quality assurance (QA) protocol ideally verifying
positioning accuracy for each patient as recommended by AAPM (Nath et al 1997). Most of
the measurements are currently performed using radiochromic films (Buzurovic et al 2016,
Granero et  al 2016) that can provide high 2D spatial resolution. Measurements performed
with C-arm were also reported in the literature (Goulet et al 2016).
Accurate measurements during the commissioning has proven to be especially relevant for
ring applicators. Awunor et al showed up to 0.61 cm deviation from the expected source dwell
position and differences up to 0.26 cm between a subset of ring applicators (Awunor et  al
2013). A recent multicenter study evaluated 86 ring applicators showing good consistency
between them, although up to 0.54 cm positional variation was measured for a standard plan
(Awunor et al 2015).
Although film measurements can be accurate, the time necessary to perform several meas-
urements can be significant as a new film should be positioned before each experiment. This
increases positioning uncertainties due to the placement of the films on the holder and align-
ment of the applicator on top of the films. Therefore, the time to align the films, process the
results and also the cost of the films limit the number of measurements. In addition, it is not
possible to verify short interdwell distances (⩽0.5 cm for a ring applicator using single meas-
urements) commonly employed clinically since the film exposures of the dwell positions can
partially overlap and the integrated response of the film is not sufficient to visually distinguish
the dwell positions.
In this work we describe the development of a novel method (patent pending) to measure
dwell positions and dwell times using an imaging panel (IP) that has the same capabilities as
radiochromic film and additionally provides time-resolved results allowing the instantaneous
evaluation of each dwell position individually. The main goal of this work is to describe the
developed method showing its applications, relevance in the clinic and main advantages com-
pared to the current practice. We do not aim to show the whole commissioning of an applicator
nor compare multiple applicators.

2.  Materials and methods

This section  describes the acquisition of an image of a ring applicator, using 192Ir gamma
photons, and dwell position and dwell time measurements performed using an IP. All the
measurements were performed using a GammaMed Plus iX afterloader (Varian Medical

8361
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

Systems, Palo Alto, CA) equipped with a GammaMed Plus HDR 192Ir source (3.5 mm long
active core) (Ballester et al 2001) and a ring applicator (3D Interstitial Ring Applicator, 60°
SET—GM11010190; Varian Medical Systems, Palo Alto, CA) employed for 192Ir high dose
rate (HDR) gynaecological treatments. The applicator has two channels, one for the ring (3 cm
diameter) and another for the tandem. However, only the ring commissioning is described in
this work since it is the most complex part of the applicator and the commissioning of the
tandem follows the same method (the tandem would be evaluated separately from the ring and
positioned horizontally on the surface of the holder). Two new applicators of the same model
(expected to be identical) were used during the experiments. The first applicator was used for
all the measurements except for the absolute position measurements. It was replaced due to an
intermittent problem (described in the results section) so the absolute position measurements
were performed with a second applicator of the same model.
The length of the channel (transfer tube  +  ring applicator) is 130 cm, which is the most
distal dwell position and represents the tip of the applicator. All the offsets described in this
manuscript use the tip as reference. Results are shown either using the linear length (dwell
position related to the length of the channel) or 2D Cartesian coordinates measured with
the IP.
Dwell time measurements show no more than 0.2 s difference from expected values (com-
patible with the acquisition rate of the panel (≈0.14 s per frame) and are not described in this
study, since dwell times measured with the same panel and source are part of a previous study
(Fonseca et al 2017).

2.1.  Imaging panel—IP

The measurements were performed using an IP, model XRD 1640 AG9 ES (Perkin Elmer,
Waltham, MA), with an amorphous silicon sensor operating as a 2D array of 40.96  ×  40.96 cm2
(1024  ×  1024 pixels) and pixel size of 0.04 cm (Podesta et al 2012). The maximum acquisi-
tion rate of the panel is approximately 7 fps (0.14 s per frame) with frames transmitted directly
to a computer connected to the panel. Therefore, the dwell time should be more than 1 s to
assure that at least a few frames are acquired with the 192Ir source at the dwell position and to
exclude the frames acquired with the source moving in between the dwell positions. Results
were processed frame by frame to identify dwell position and measure dwell times.
IP results were validated against film measurements with most of the points showing agree-
ment within 0.3 mm as described in a previous study, which also describes the calibration of
the panel and dwell time measurements (Fonseca et al 2017). Image corrections due to pixel
sensitivity, energy and angular dependence (McCurdy et  al 2001, McDermott et  al 2004,
Smith et al 2013) have a negligible impact in the results described in this manuscript due to
the small distance between the source and the panel resulting in an intense and well defined
peak. The most relevant correction is the subtraction of the dark field (image acquired without
radiation), which was applied to all the frames.

2.2.  Experimental setup

2.2.1.  192Ir imaging.  Figure 1 shows a sketch of the experimental setup (an illustration of
the method can be seen in video 1). The IP measurement consists of three steps: (1) back-
ground (BG) acquisition (figure 2(a)); (2) acquisition of an image of the applicator (figure
2(b)); (3) dwell position measurements (figure 2(c)). The first two steps are necessary to
acquire an image of the applicator using gamma photons emitted by the 192Ir source. The BG

8362
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

Figure 1. Representation of the experimental setup for imaging a brachytherapy

applicator with an 192Ir source.

Figure 2.  (a) BG acquisition without including the applicator; (b) image acquisition
with the applicator in place; (c) dwell position verification sending the source inside
the applicator. The solid red arrows point to the source position during the image
acquisition and the hollow red arrow points to the position of the applicator during the
dwell position measurements.

measurement acquires an image without the applicator and is necessary to remove the effect
of the holder, double tape, pixel sensitivity differences along the IP and the PMMA plate on
the top of the IP. In addition, the BG image corrects for the point source effect that results in
high intensity regions right below the source with the response decreasing with the distance.
The ring applicator was fixed, using double-sided tape, at the surface of a 30  ×  30  ×  1 cm
PMMA plate that was fixed on top of the IP to create some distance between the source and
the IP to avoid response saturation. An image was acquired with the applicator in place follow-
ing exactly the same process for the BG acquisition. The reference image was then obtained
by subtracting the BG from the image obtained with the applicator in place. The BG and appli-
cator images were acquired integrating the IP response over 60 s with the source positioned
20 cm above the center of the surface of the PMMA plate. In addition, BG and applicator
images were acquired with the source positioned 35 cm above the surface of the PMMA plate
for comparative purposes. The images acquired for the second applicator used in this work
included radiopaque markers and a dummy wire inserted into the applicator to visualize the
source trajectory and define the tip of the catheter.

8363
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

The acquired image corresponds to a projection of the applicator since we used an approxi-
mate point source and the applicator is not directly on top of the active layer of the panel as
illustrated in figure 1. The active layer of the IP is approximately 1 cm below its surface. The
position of the active layer was defined using the technical specifications of the panel and meas-
urements using a reference object with known dimensions (lead block with 14  ×  14  ×  2 cm)
on the top of the panel (data not shown).
The projection of the applicator acquired at the active layer depth was corrected to rep-
resent a projection at the center of the ring (2.5 cm above the active layer). The source posi-
tion (xy—measurement plane) during the image acquisition was defined using the BG image
where the high intensity region below the source is clearly visible, and by employing a 2D
Gaussian fit (described in section 2.3). The distance between the source and the surface of the
PMMA plate was measured using a regular ruler.
Figure 1 shows an arbitrary point P and its representation (P′) in the projected image. The
distance (d′) between P′ and the projected source position at the level of the active layer is
larger than the distance (d) between P and the projected source position at the level of the
applicator. This effect was corrected using the relation d /h = d/h to calculate the displace-
ment of P. The described process was repeated for all the pixels (xy direction) applying bilin-
ear interpolation when necessary.

2.2.2.  Dwell position measurements.  All the experiments were performed without moving
the applicator between the acquisition of the applicator image and the irradiation within the
ring applicator. Several measurements were performed to define the dwell positions and ver-
ify their reproducibility as described below. Dwell positions were compared from repeated
measurements to assess reproducibility and against each other to define interdwell distances.
Absolute measurements in relation to the tip of channel are described in section 2.2.2.4.

2.2.2.1. Interdwell distances.  Interdwell distances of 1.0 cm, 0.5 cm, 0.2 cm and 0.1 cm were
measured twice.

2.2.2.2. Inclination of the applicator and transfer tube curvature.  Measurements with 0.5 cm
interdwell distance (first position at 129.5 cm) were performed for 3 different inclinations of
the IP and applicator with respect to the horizontal (0° (figure 2(c)), ≈45° (figure 3(a)) and
90° (figure 3(b))). Experiments were performed with the transfer tube straight and at the same
height as the connection of the applicator (recommended by the manufacturer) (figure 3(b))
and with increased curvature of the transfer tube (figure 3(a)) for all the inclinations. Two
measurements were performed for each situation, resulting in 12 measurements in total.

2.2.2.3. Reproducibility.  The reproducibility was verified by performing 10 consecutive mea-

surements with 7 dwell positions (interdwell distance of 0.5 cm) and 3 s dwell time for each
position. The experiments were performed for two different panel/applicator inclinations
(0°—figure 2(c), and 90°—figure 3(b)) using both the transfer tube straight and curved total-
ing 40 measurements.

2.2.2.4. Absolute positions.  The measurements described in this section  were performed
using a second applicator of the same model since the applicator used for the measurements
described in the previous sections had to be replaced (due to an intermittent problem described
in the results section). A new reference image of the applicator was obtained following the
same procedure described above. In addition, images were acquired using radiopaque mark-
ers, inserted up to the tip of the channel, and a dummy source, manually inserted, at the tip and

8364
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

Figure 3. (a) Measurement performed with the panel at ~45° with respect to the
horizontal and a curved transfer tube. (b) Measurement performed with the IP at 90°
with respect to the horizontal and the transfer tube aligned with the applicator. The
experiments were repeated, for all the inclinations, using the transfer tube aligned and
curved. The arrow indicate the positions of the applicator.

1.6, 3.1, 4.5, 6.0, 7.4 and 8.2 cm from the tip. These images were acquired to define the tip of
the channel (reference for the treatment planning system) and to evaluate the trajectory of the
source since it can move between the inner and outer walls of the channel because its diameter
is larger than the diameter of the source cable. Measurements were performed for 0.2 cm (43
dwell positions) and 0.5 cm (18 dwell positions) interdwell distances using an offset of 0.5 cm.
The measurement was repeated 10 times for each interdwell distance.

2.3.  Data analysis—IP

The IP acquires around 7 fps totaling thousands of frames for the described experiments,
which cannot be processed manually. Software was developed using MatLab, version 8.5
(The MathWorks, Inc., Natick, MA) to process the frames defining the source position ( x0 , y0 )
using a 2D Gaussian fit (equation (1)), where A is the height of the peak.
    
2 2
(x − x0 ) (y − y0 )
F (x, y) = A ∗ exp −
(1) − .
2σx2 2σy2

The frames were corrected by subtracting the dark field (response of the IP without any
radiation). A pre-defined threshold of 0.03 cm and 3 frames was employed to define different
dwell positions. Therefore, if the source stays at the same position (within 0.03 cm) for at least
3 consecutive frames the software identifies a dwell position.
The software, IrIS (Iridium Imaging System), has two outputs: a table showing all the dwell
positions and their absolute coordinates in centimeters; and a video showing the source move-
ment inside the applicator. The images acquired during the irradiation with the 192Ir source
inside the applicator were overlapped with the gamma-ray image of the applicator allowing a
visual inspection of the results. See the supplementary material for output examples (stacks.
iop.org/PMB/62/8360/mmedia). IrIS runs as a standalone package without Matlab.
The interdwell distances described in this work consist of the linear distance between two
consecutive dwell positions. Therefore, the interdwell distance does not account for the cur-
vature of the applicator. It is a chord of a circumference and not the arc (real trajectory of the
source). However, the difference between chord and arc lengths is negligible for the interd-
well distances measured in this work (up to 1 cm). The absolute distance in relation to the tip
of the applicator uses the arc length (real trajectory) since large distances (up to 9 cm) were
measured.

8365
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

Figure 4.  IrIS screenshot showing an image of a ring applicator, acquired using an 192Ir
source, for dwell positions (green (first measured dwell positon) and red dots) and an
overlaid IP frame showing a dwell position (full arrow). The user can set contrast and
transparency options to improve visualization.

3. Results

All the data obtained with the IP was processed using IrIS which corrects the projected dimen-
sions of the applicator, calculates the dwell positions and provides several visualization tools.
Figure  4 shows a screenshot of the software that is also capable of controlling the IP and
processing the data while the IP is still acquiring the images. Therefore, the result analysis is
concluded shortly after the irradiation (~1–2 min). Video 2 shows an example.

3.1.  Image acquisition using an 192Ir source

Figure 5(a) shows the BG image acquired without the applicator, showing a high intensity
region at the center of image, corresponding to the region right beneath the source position.
The applicator is clearly visible in figure 5(b), although the image can be much improved by
subtracting the BG (figure 5(c)). The image is then corrected to account for the influence of
the diverging source gamma rays (assuming an isotropic point source) and depth of the active
layer (figure 5(d)). The difference in size can be significant when the image is acquired with
the source close to the panel (20.0 cm in this case) as can be seen in figure 5(e) where the
acquired projection (gray) is overlaid with the corrected applicator image (yellow–red). The
corrected image acquired with the source at 20.0 cm (yellow–red) and 35.0 cm (grey) above
the PMMA plate show a very good agreement in the region of the dwell positions (figure 5(f)).

3.2.  Interdwell distances

Figure 6(a) shows the results for 0.5 cm interdwell distance. The first interdwell distance
is 0.26 cm; however, this is a known issue and the manufacturer recommends an offset of
5 mm so the first dwell position (white dot at 130 cm) should not be used and the measured

8366
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

Figure 5.  (a) BG image acquired without the applicator; (b) image acquired with the
applicator in place; (c) projection of the applicator corrected by the BG; (d) image
of the applicator corrected by the point source and sensitive layer depth effects; (e)
projected image without correction (grey) on top of the corrected image (yellow–red);
(f) corrected images obtained with the source at 20.0 cm (yellow–red) and 35.0 cm
(grey) from the IP. The arrow indicates a region below the cable of the applicator that
does not show a good agreement since this part of the applicator is not directly on the
surface of the PMMA (figure 1). This is irrelevant since there are no dwell positions in
this region.

8367
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

Figure 6. (a) IP measurements for 0.5 cm interdwell distance between 130.0 and
121.5 cm. (b) Two series (circles and triangles) of consecutive IP measurements for
0.2 cm interdwell distance between 130.0 and 121.2 cm. (c) Magnification of (b)
highlighting the region for which the largest interdwell distance was measured (red
arrow). The white dot represents the first dwell position with the source moving in
counter clockwise direction.

difference can be ignored. The remaining values are between 0.44 cm (7th interdwell dis-
tance) and 0.54 cm (18th interdwell distance). The measurement for 0.2 cm interdwell distance
shown in figure 6(b) (magnified in figure 6(c)) shows a 0.4 cm measured interdwell distance
between the positions 128.4 and 128.0 cm (indicated in figure 6(c)). The measured dwell times
show that the source stays in position128.4 cm for twice the programmed dwell time and then
moves to the position 128.0 cm skipping the position 128.2 cm (figure 6(c)).
Table 1 shows the measurements for interdwell distances between 0.1 and 1.0 cm. The first
dwell position (130 cm) was ignored since an offset would be applied for clinical cases.

8368
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

3.3.  Inclination and transfer tube curvature effects

The curvature of the transfer tube can result in differences larger than 0.1 cm in dwell posi-
tions as it can change the length of the trajectory followed by the source. Figure 7 shows the
difference between the positions obtained with the catheter aligned and curved (for the same
inclination of the panel/applicator, 0–90°) and a comparison between the results obtained
with the catheter aligned for different inclinations of the panel/applicator. The effect of the
curvature seems to be less relevant when the applicator is inclined (90°), although there is no
obvious explanation since results obtained with both inclinations (45° and 90°) do not show
significant differences when compared to the results obtained without inclining the applicator/
panel (0°). The experiment with 90° inclination and a curved transfer tube shows a 0.30 cm
difference at the position 128 cm due to the intermittent problem described in the next section.

3.4. Reproducibility

Figure 8 shows the results obtained for 2 different inclinations (0° and 90°) with the trans-
fer tube aligned and curved. Ten measurements were performed for each situation without
moving the afterloader. The horizontal axis shows the programmed dwell positions related
to the tip of the applicator and the vertical axis shows the standard deviation for each series
separately and for all the data combined (points connected with dashed lines for visualization
purposes) for x and y coordinates. The dwell positions are very reproducible with standard
deviation  <0.03 cm for most of the positions. The standard deviation increases combining all
the measurements due to the different curvature of the transfer tube and repositioning of the
afterloader, although values are  <0.05 cm, except for one position. The largest differences
were observed for the position 128 cm with a standard deviation of 0.17 cm. An intermittent
problem was observed at the position 128 cm with some measurements showing a very small
displacement between the positions 128.5 and 128 cm. The problem was observed in the same
region of the applicator where one dwell position was skipped using 2 mm interdwell distance
(figure 6(c))

3.5.  Absolute reference

The absolute coordinates are defined using the tip as reference which must be accurately
defined. Figure 9(a) shows an image of the applicator with radiopaque markers. It is very dif-
ficult to define the tip of the channel since there is an overlap, in the projected image, between
the entrance of the ring and the end of the channel (figure 9(a)—solid arrow). There is no
real overlap since the entrance of the ring and the tip are at different heights, which cannot
be distinguished using a 2D projection. The radiopaque markers are close to the outer edge
of the channel (figure 9(a)) whilst the dwell positions (figure 6) are closer to the center of the
channel. Figure  9(b) shows that the dummy wire overlaps with the markers. However, the
images acquired retracting the dummy wire shows that the tip does not completely overlap
with the radiopaque markers and the center of the source is closer to the center of the channel.
It indicates that the center of the channel (figure 9(a)—hollow arrow) should be used to define
the contour of the channel.
The image of the applicator with the dummy wire (figure 10(a)) shows the same overlap
observed using the markers. However, it was possible to define the tip of the channel artifi-
cially by removing the applicator from the image to improve the visualization of the dummy
wire (figure 10(b)). The highest intensity region shows the overlap region. Figure 10(c) shows
the contour of the channel whose tip was defined using figure 10(b) and the remaining points

8369
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

Table 1. Measurements performed with an IP for interdwell distances between 0.1

and 1.0 cm showing the mean (±1σ), minimum and maximum measured values. Two
consecutive measurements were performed for each interdwell distance. The values
in brackets correspond to the maximum difference excluding the point that differs the
most from the expected value for 0.1 and 0.2 cm interdwell distances.
Interdwell distance IP (cm)
Expected value Number of
(cm) Mean  ±  σ Minimum Maximum dwell positions
0.1 0.10  ±  0.04 0.04 0.30 (0.21) 108
0.2 0.20  ±  0.04 0.13 0.42 (0.23) 80
0.5 0.49  ±  0.04 0.44 0.56 40
1.0 0.97  ±  0.06 0.90 1.07 18

Figure 7.  Difference between the expected dwell position (horizontal axis) and the
measured value for the transfer tube aligned and curved for the same IP inclination
(0, 45 and 90°) and different IP inclinations keeping the transfer tube aligned. Dashed
lines connecting the points were added to guide the eye and highlight the results
comparing different inclinations.

defined following the center of the channel (figure 10(c)). The red circles are measured dwell
positions starting at 129.5 cm with 0.5 cm interdwell distance. The measured dwell positions
were added to verify the contour of the channel.
Figure 11 shows the differences between the programmed and measured dwell positions
for interdwell distances of 0.2 and 0.5 cm (starting at 129.5 cm). The values correspond to the
position in relation to the tip of the channel (130 cm). The differences goes from  −0.10 up to
0.18 cm in accordance with the 0.2 cm accuracy described by the manufacturer.

4. Discussion

Awunor et al described an expanded measurement uncertainty (two standard deviations) of

0.14 cm obtained using a custom made holder and markers in a multicentric study totaling
more than 800 film irradiations. The measurements were performed in two consecutive runs to
8370
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

Figure 8.  Standard deviation measured for two panel/applicator inclinations with the
transfer tube aligned and curved. Ten measurements were performed for each situation
totaling 40 measurements. The results for all the measurement combined are connected
with dashed lines for visualization purposes.

minimize the response overlap resulting in a blurred film. Therefore, 1 cm interdwell distances
were verified in two different irradiations with 2 cm interdwell distances adding a 1 cm offset
to the second measurement (Awunor et al 2015). In theory, the experimental setup can always
be improved, although uncertainties lower than described by Awunor et al (2015) would be
very difficult to achieve using films. In addition, short interdwell distances cannot be mea-
sured in a single irradiation, which increases the uncertainty and may lead to measurements
not exactly reproducing the clinical situation.
Time-resolved measurements overcome the overlap limitation allowing measurements of
0.1 cm interdwell distance in a single irradiation. The uncertainty of the dwell position using
the IP is less than 0.02 cm (type A—1σ) due to the 2D Gaussian fit that is very robust with
uncertainties smaller than the pixel size of the panel (0.04 cm). The IP frame rate used is not
high enough to measure precisely the source position during its movement since the IP is read
out line by line with a small delay between them resulting in a distorted response. This effect
was ignored since the goal of this work is to verify the dwell positions. A panel with a higher
frame rate may enable tracking a moving source.
The IP used in this work was removed from a LINAC after few years of use and has been
employed for the measurements described in this paper for at least 11 months without show-
ing any observable reponse or performance variation. However, we intend to implement peri-
odical checks after source replacements (~3 months) until more data is available about the IP
aging effects.
The image of the applicator acquired using the IP needs to be corrected due to the depth
of the active layer and point source effects. However, the correction is very simple and can
be easily verified comparing the dimensions of the corrected image against the manufacturer
specification. In addition, the video output of IrIS shows the source position (IP images with-
out corrections) that must follow the channel of the applicator. During the image acquisition

8371
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

Figure 9.  (a) Image of the applicator with radiopaque markers inserted in an attempt to
identify the tip of the catheter. The solid arrow indicates the region where the entrance
and the tip of the catheter overlap in the projected image. The overlap happens since
the entrance of the channel of the ring is at a different height so the tip of the channel is
beneath the entrance of the ring. The hollow arrow indicates the center of the channel;
(b) image of a dummy wire, acquired with the 192Ir source, and the IP overlaid to (a);
(c) same as previous with the dummy wire at 3.1 cm from the tip; (d) same as previous
with the dummy wire at 6.0 cm from the tip.

the source should be placed about 20 cm above the applicator so that the uncertainty of the
height (±0.05 cm) has a very small effect. Assuming the source was placed approximately
20 cm above the center of the applicator a 0.2 cm difference in the height would result in a dif-
ference of less than 0.01 cm in the applicator projection. The image of the applicator obtained
with the source at 20 and 35 cm above of the applicator is very similar so 20 cm is preferable
to improve image quality due to the higher IP response. The projection of the applicator and
the dummy wire was essential to define the end of the channel that is used as reference in the
treatment planning system. The possibility to subtract the BG and remove the applicator leav-
ing only the dummy wire greatly improves the visualization.

8372
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

Figure 10.  (a) Image of the applicator with a dummy wire. The arrow indicates the
region where the entrance of the ring applicator and the tip of the channel overlap in
the projected image. (b) Enhanced image of the dummy wired obtained by subtracting
the image of the applicator (without the dummy) from the image of the applicator with
the dummy wire. The arrow shows the highest intensity region due to the overlap.
Intensity values were inverted so the region with the overlap (more attenuation) became
the brightest region. (c) Image of the applicator with the contour of the channel (green
line) and red circles showing measured dwell positions (starting at 129.5 cm with 0.5 cm
interdwell distance) included to verify the contour of the channel.

Figure 11. Difference between the programmed and measured dwell positions for
interdwell distances of 0.2 and 0.5 cm with the transfer tube aligned. The measured
values are the mean of 10 consecutive measurements with the error bars representing
one standard deviation.

Awunor et al (2013) and Hellebust et al (2010) highlighted the importance of individually
commissioning each ring before acceptance for clinical use due to the observation of some
outliers. We also recommend individual commissioning and encourage the clinical staff to
perform several measurements for the same applicator, which is easy to do with the system
proposed here.

8373
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

An intermittent problem was observed for one applicator used in this work. The frequency
of the problem reduced over time going from 100% during the first experiments to  ≈10% after
several measurements. Therefore, 2 or 3 measurements may not be enough depending on the
frequency of the problem. The reason of the problem is still not clear, although we believe the
source was intermittently getting stuck between the position 128.5 and 128.0 cm. It is possible
the source gradually sanded down the surface of the applicator reducing the frequency of the
problem after each irradiation. Although the problem was not in a dwell position expected to
be actively used in clinical treatments, the applicator was replaced to completely eliminate the
risk of use of a wrong dwell position (we did not observe the same problem with the second
applicator). Besides the problem between the position at 128.5–128.0 cm, the measurements
were very reproducible showing a standard deviation below 0.03 cm for most of the points
obtained with the same transfer tube curvature.
Differences are larger at certain positions (figure 11) of the applicator most likely due to the
trajectory of the source measured in 2D, whilst the diameter of the channel allows small move-
ments in 3D. The internal diameter of the applicator is larger than the diameter of the source
allowing for the curvature of the source cable, which can lead to small positioning variations
within the channel. It might be the case the source follows a trajectory close to the bottom at
certain regions of the applicator and close to the top in another region.
The IrIS software developed allows visual inspection of the results by generating mov-
ies showing an image of the applicator, the movement of the source and the identified dwell
positions in real time. It is a very important feature to identify errors since all the steps of the
irradiation can be visualized and recorded for future reference.

5. Conclusions

The IP and the IrIS software overcome the film limitations for brachytherapy applicator veri-
fication and provide several additional advantages. A projection of the applicator can easily be
obtained using the 192Ir gamma photons without requiring any additional source of radiation,
e.g. an x-ray device. Measurement times are drastically reduced since several irradiations can
be performed sequentially without replacing the films, which can be very time consuming
since usually reference marks need to be drawn using a pen on each film. Measurement uncer-
tainties are much smaller using IP, which allows the verification of the smallest interdwell
distance available (0.1 cm). The IP system introduced here can also be used to verify applica-
tors, other than ring applicators. The initial cost of acquiring an IP may be higher than the cost
of films. However, we believe that the advantages of the method easily justify its cost and that
disposable films can be even more expensive than an IP in the long term. In addition, the IP
reduces personal cost since it reduces measurement time.

References

Awunor  O, Berger  D and Kirisits  C 2015 A multicenter study to quantify systematic variations and
associated uncertainties in source positioning with commonly used HDR afterloaders and ring
applicators for the treatment of cervical carcinomas Med. Phys. 42 4472–83
Awunor O A, Dixon B and Walker C 2013 Direct reconstruction and associated uncertainties of 192Ir
source dwell positions in ring applicators using gafchromic film in the treatment planning of HDR
brachytherapy cervix patients Phys. Med. Biol. 58 3207–25
Ballester F, Puchades V, Lluch J L, Serrano-Andrés M A, Limami Y, Pérez-Calatayud J and Casal E
2001 Technical note: Monte-Carlo dosimetry of the HDR 12i and Plus 192Ir sources Med. Phys.
28 2586–91

8374
Phys. Med. Biol. 62 (2017) 8360 G P Fonseca et al

Buzurovic  I  M, Hansen  J  L, Bhagwat  M  S, O’Farrell  D  A, Friesen  S, Harris  T  C, Damato  A  L,

Cormack R A, Martin N E and Devlin P M 2016 Clinical implementation of a novel applicator in
high-dose-rate brachytherapy treatment of esophageal cancer J. Contemp. Brachyther. 8 319–25
Fonseca  G  P, Podesta  M, Bellezzo  M, Bosch  M  R  V  D, Lutgens  L, Vanneste  B  G  L, Voncken  R,
Limbergen E J V, Reniers B and Verhaegen F 2017 Online pretreatment verification of high-dose
rate brachytherapy using an imaging panel Phys. Med. Biol. 62 5440
Goulet  M, Lacroix  F and Aubin  S 2016 Gynecological ring applicator: three dimensional dwell
positioning accuracy, long-term stability, angular accuracy and their impact on delivered dose
Brachytherapy 15 S116–7
Granero  D, Candela-Juan  C, Ballester  F, Ouhib  Z, Vijande  J, Richart  J and Perez-Calatayud  J 2016
Commissioning and quality assurance procedures for the HDR Valencia skin applicators
J. Contemp. Brachyther. 8 441–7
Hellebust T P et al 2010 Recommendations from gynaecological (GYN) GEC-ESTRO Working Group:
considerations and pitfalls in commissioning and applicator reconstruction in 3D image-based
treatment planning of cervix cancer brachytherapy Radiother. Oncol. 96 153–60
McCurdy B M, Luchka K and Pistorius S 2001 Dosimetric investigation and portal dose image prediction
using an amorphous silicon electronic portal imaging device Med. Phys. 28 911–24
McDermott  L  N, Louwe  R  J, Sonke  J  J, van Herk  M  B and Mijnheer  B  J 2004 Dose-response and
ghosting effects of an amorphous silicon electronic portal imaging device Med. Phys. 31 285–95
Nath R, Anderson L L, Meli J A, Olch A J, Stitt J A and Williamson J F 1997 Code of practice for
brachytherapy physics: report of the AAPM Radiation Therapy Committee Task Group No. 56.
American Association of Physicists in Medicine Med. Phys. 24 1557–98
Podesta M, Nijsten S M, Snaith J, Orlandini M, Lustberg T, Emans D, Aland T and Verhaegen F 2012
Measured versus simulated portal images for low MU fields on three accelerator types: possible
consequences for 2D portal dosimetry Med. Phys. 39 7470–9
Smith R L, Taylor M L, McDermott L N, Haworth A, Millar J L and Franich R D 2013 Source position
verification and dosimetry in HDR brachytherapy using an EPID Med. Phys. 40 111706

8375