Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
SCIENTIFIC NOTE
Practical IMRT QA dosimetry using Gafchromic film:a quick start guide
Nick Bennie1,2 • Peter Metcalfe2
Received: 26 August 2015 / Accepted: 5 April 2016 / Published online: 20 April 2016
� Australasian College of Physical Scientists and Engineers in Medicine 2016
Abstract This work outlines a method for using Gaf-
chromic film for dosimetry purposes, by scanning it with
currently available commercial scanners. The scanners
used were: Epson V800, Epson V700, Epson V37 series,
specifically a V370 and a Canon multi-function office
printer/scanner. The Epson scanners have 16 bit RGB
resolution, the Canon has 8 bit RGB (Red Green Blue)
resolution, and the V800 and V700 allow scanning in
transmission mode. The V700 uses an Epson White Cold
Cathode Florescent Lamp; the recently released V800 uses
an Epson light emitting diode (LED) light source, while the
V37 series uses a reflective mode and the Epson LED light
source. The Epson V37 series scanners are designed for
non-professional use so the cost has been kept at a low
‘‘entry level’’ point, so they would be a suitable option for a
department wanting to use Gafchromic film or with limited
needs that did not justify a more sophisticated and expen-
sive unit. Note that the V800 or V700 scanners are not
expensive in context, costing approximately the same as a
25 sheet box of Gafchromic film. The Canon was included
to demonstrate that a scanner with 8 bit RGB resolution can
be used for dosimetry. These general multi-function units
are available in most departments, and they would allow
Gafchromic film to be evaluated as a dosimetry tool
without a significant investment. Furthermore, they are
generally capable of scanning large format film
(425 9 350 mm) in one part. Although this is not neces-
sary for dosimetry, it is often useful for machine QA,
where dividing the film into two parts to ensure accurate
measurements is not practical. Moreover, this analytical
method uses software that is freely or commonly available,
particularly the image processing package ImageJ. Note
ImageJ v1.48 was the version used. The results demon-
strate that this method used with the scanners evaluated is a
practical method of using Gafchromic film as a dosimeter
for IMRT QA.
Keywords Gafchromic � LED light source � IMRT QA �ImageJ
Introduction
The North Coast Cancer Institute (NCCI), like many other
radiotherapy based physics departments, has a need for film
dosimetry, but they no longer have access to traditional wet
film, silver halide, and processing facilities so it was
decided to use Gafchromic radiochromic film [1] for
dosimetry and for assessing dosimetry in the transverse
plane of IMRT (Intensity Modulated Radiotherapy) QA
(Quality Assurance) studies. Gafchromic radiochromic film
has been considered difficult to work with, this may be
because it is treated as a direct replacement for traditional
wet processed, silver halide, film, whereas it has charac-
teristics that are different. The use of commercially
Presented as a Poster at EPSM 2012, The 36th Annual Engineering
and Physical Sciences in Medicine conference, Gold Coast, Australia,
2–6 December, 2012.
& Nick Bennie
Peter Metcalfe
1 Lismore Base Hospital, NCCI, Northern NSW Local Health
District, Lismore, NSW 2480, Australia
2 Centre For Medical Radiation Physics, University of
Wollongong, Northfield Avenue, Wollongong, NSW 2500,
Australia
123
Australas Phys Eng Sci Med (2016) 39:533–545
DOI 10.1007/s13246-016-0443-0
available flatbed scanners to scan this film further differ-
entiates it from the traditional methods used with silver
halide film.
General purpose commercial scanners are not optimised
for scanning Gafchromic film for dosimetric purposes
because the response of the scanner in the axis perpen-
dicular to the scan direction is non-uniform; this means the
derived doses vary more than the generally accepted tol-
erances for IMRT QA analysis [2–5].
Furthermore, the active layer of Gafchromic film is not
deposited uniformly, which may result in discrepancies in
derived doses that are greater than acceptable [6, 7]. Sev-
eral methods have been proposed to deal with this
including: Micke et al. [6–11].
The method presented here demonstrates a simple and
practical method using tools from a standard image pro-
cessing application, ImageJ Rasband, NIH [12]. The use of
standard tools may allow the reader to easily understand
and implement this method in their department [13].
The intention is to allow a user to fully understand
each step in the process. It demonstrates that scanners that
are generally available and software that is free or com-
monly available can be used for IMRT dosimetry. In
general, this would allow a department to evaluate Gaf-
chromic film for dosimetry without the need to purchase
additional equipment.
The method demonstrates that a function can be derived
(derived function) from the scanner film combination that
is approximately directly proportional with dose.
It is generally recommended that a scanner with 16 bit
resolution in transmission mode, such as the Epson V700
or V800 be used. An assessment of the Epson V37, which
works in reflective mode, indicated it was good, while the 8
bit resolution Canon (multi-function printer/scanner, model
designation for specific unit used Canon iRC3580i) which
works in reflective mode was assessed as acceptable.
Methods
The method chosen used a scan of the unexposed film as a
reference (correction map) because it is simple to under-
stand and implement, and it leads to a derived value as a
first approximation that is directly proportional to dose. As
such it is valid for IMRT QA for exposures of up to 50 Gy.
Note that this is a system correction that corrects the factors
from the scanner and the factors from the film.
Although the 3 channel method proposed by Gaf-
chromic Micke et al. [5] does not need an unexposed scan,
the processing method is complicated and when done by
proprietary software, the system is considered to be diffi-
cult to understand. Moreover, there is a significant cost
involved with this method which a department would need
to consider carefully before purchase. It was also noted that
Gafchromic do provide this software for evaluation on
request.
Theory
In general, and specifically for the scanners used, the
system response (pixel value) with respect to an absorbed
dose at a specified point of the film, and for a single
channel of a RGB scan, was modelled by a 3 term
hyperbolic function; see the Gafchromic whitepaper Gaf-
chromicVR [1].
PxðDÞ ¼ ðaþ bDÞðcþ DÞ
where PxðDÞ is the raw pixel value from the scan, D is the
dose the pixel has been exposed to, a, b, c which are
constants, then for an unexposed ‘‘Blank’’ film, D = 0,
giving
Pxð0Þ ¼ a
c
For a saturated film exposed to a very high dose, Limit as D
??
Pxð1Þ ¼ b
If b and c are small then this function may be approxi-
mated by a simple hyperbolic such that when the unexposed
‘‘Blank’’ is divided by the ‘‘Exposed’’ scan, the result is a
function that is approximately linear for a small c and b
Pxð0ÞPxðDÞ ffi 1þ D
c
Then the function is derived further:
Pxð0ÞPxðDÞ � 1 ffi D
c
So that it is directly proportional to dose, D.
For the Green channel of RGB data this relationship
holds reasonably well for most scanners.
For the Red channel it appears that the unexposed
‘‘Blank’’ was unstable. With regards to introduced noise, it
appears that more noise was introduced than could gener-
ally be compensated for. It appears the ‘‘Blank’’ is almost
transparent to red, which means the measured values are
near a limit of the amplitude resolution of the scanner,
leading to non-linear or truncation effects, resulting in
significant noise and scan to scan inconsistency.
For the Blue channel, the system sensitivity was notably
less than the Red or Green channels. It was noted that the
Gafchromic marker dye is active in the Blue channel.
Therefore the Green channel was considered the most
suitable to use with this method.
534 Australas Phys Eng Sci Med (2016) 39:533–545
123
Recipe and hints for using different scanners
A sheet of Gafchromic film should be selected.
The film should be marked such that the orientation can
be replicated, while any other marks may be added for
identification and setup, e.g., to locate the treatment iso-
centre.
Note that a standard uncut piece of film is 254 mm 9
203 mm.
It is important to preserve the orientation of the film
from Blank scan to Exposed scan, particularly with EBT2,
because there is a significant change in system response if
the film is rotated. The orientation from Blank scan to
Exposed scan must be consistent.
Using the scanner
Acquire an unexposed scan of the piece of film to be
used (Blank).
Install the film in a phantom and expose it to IMRT
treatment.
Wait for approximately 1 h.
Acquire a scan of the exposed film. (Exposed)
Store scan files in the working folder of a workstation
computer.
Using Image processing software (ImageJ):
Convert 3 channel RGB image to 3 images, one for each
channel R,G, and B.
Select the image that represents the Green channel. The
other data are not required.
Convert data to 32 bit single precision format. This
preserves resolution during processing.
Apply an Average filter to both Blank and Exposed. This
enhances amplitude resolution.
Divide scan of Blank by scan of Exposed.
Subtract unity (1) from all pixels of image.
Apply Median filter to remove noise enhanced by
division process.
Multiply result by constant of proportionality (derivation
follows below). Result is now in units of Dose.
Use Resize function to convert image pixel resolution to
the dimensions of the active area in millimetres.
254 mm 9 203 mm if using standard size uncut film.
Use Minimum function to set all values less than 0 to 0.
Use the Maximum function to limit the highest value.
High values can be introduced by noise or markers on
the film, and they may cause confusion in an IMRT
comparison package such as SNC Patient (Sun Nuclear
Corporation). For example, if the planned data indicated
the maximum value was 250 cGy it would be reasonable
to limit the maximum of the film data to 300.
This data is referred to as Processed Scaled Measured Data
(PSMD).
Save the PSMD as an image in ‘‘Text Image’’ format.
Data in ‘‘Text Image’’ format can be easily converted to
a format that is recognised by IMRT QA analysis programs
such as SNC Patient. See ‘‘Appendix 2’’ section.
The dose distribution derived from film measurement
can be compared with the dose plane exported from the
Treatment Planning System (TPS).
There may be a small difference in the amplitude scale
that can be corrected by selecting the Relative comparison
mode of the IMRT analysis package.
Derivation of constant of proportionality
Select a pixel of film that was exposed to 2 Gy and derive
the processed measured value. The numerical value that
multiplies this derived value to equal 200 (200 cGy) is the
constant of proportionality. 2 Gy was chosen because it is
the most common fraction dose in radiotherapy.
Note that the pixel selected should have been exposed to
a known dose of 2 Gy. When the derived constant is
applied to other pixels, either on the same sheet of film, or
other sheets of film, exposed to a different distribution, but
prepared under the same conditions, then the pixels close in
value to the suggested calibration point at 2 Gy will be
reasonably accurately converted to dose. It is recognised
that the function is only approximately linear with dose. It
is recognised and addressed that for some scanners, par-
ticularly in the range 0–3 Gy, this will result in an IMRT
analysis not meeting general criteria.
Apply a calibration function using the calibrate function
from ImageJ
The response of some scanners to the Green channel with
dose was only approximately linear; for example, non-
linearity for the V700 between 0 and 3 Gy will result in
approximately a 5 % discrepancy, and if a constant of
proportionality is used and there is agreement at 2 Gy, then
there will be an apparent under response of 5 % at 1 Gy.
This is significant because it will result in errors in the
IMRT analysis that are more than the generally accepted
criteria. To address this, a simple calibration function can
be applied in ImageJ. There is no reason to suggest that the
active medium of Gafchromic film has a discontinuous
function with dose. Therefore if the function is described
accurately, there is no requirement for a multiple point by
point calibration. Note this correction appears to be scanner
specific. That is, it is related to the workings of the scanner
hardware or software and not to the conversion of the
active medium by absorbed dose.
The Power Function (ImageJ, y = Ax^B) is a simple
function that is suitable.
Australas Phys Eng Sci Med (2016) 39:533–545 535
123
To apply this calibration using the application ImageJ,
load the PSMD and create an image window with the same
dimensions, but with a 16 bit image type, and then use the
copy function to copy the 32 bit data to the 16 bit window.
Note that with ImageJ version 1.48v, 16 bit data must be
used for calibration and the data must be copied, do not use
the change Image Type function directly.
Note that 16 bit data is an integer so the data must be
scaled before conversion to maintain amplitude resolution.
Select the Calibrate function and then enter the fol-
lowing values for:
Measured: 0, 105, 200 and
Response: 0, 100, 200.
Select Function: Power. The data in the 16 bit window is
then calibrated and can then be saved in ‘‘Text image’’
format. It can also be easily converted to a format
suitable for use by an IMRT analysis program.
Results
Calibration- different system (scanner 1 film)
response
The scanners considered were: Epson V800, Epson V700,
and Epson V37 series, specifically, a V370 and a Canon
(Multi-function office printer/scanner).
Discussion
The graphs in Fig. 1 show that the scanners have various
system response functions. They can also be seen on the
reproduction of the scanned image. However, this may
depend on the quality of the print and the actual observer
because they are general purpose scanners optimised for
general office and photographic applications, and often
Fig. 1 Scanner calibration response for four different scanner systems a Epson V800 b Epson V700 c Epson V370 d Canon Multi-Function Unit
536 Australas Phys Eng Sci Med (2016) 39:533–545
123
with significant cost constraints. This means they are
generally used to copy a document or photograph that
when evaluated by eye, the copy seen on a computer
monitor appears to be similar to the original. In modern
dosimetry however, assessment is based on the numerical
data evaluated in a computer system, evaluation is almost
never done by eye. Gafchromic film is not optimised to the
human visual system, such as certain silver halide films
were, so direct visual assessment should not be attempted
without careful consideration. The spectrum output of the
light source and spectral response of the detectors used in
these systems were not optimised for Gafchromic film.
For each scanner considered the system response of the
Green channel was a well behaved function and for most, it
was approximately linear with dose. With the V800, the
response was close to linear, from 0 to greater than 50 Gy
Fig. 2. The specific definition being the error associated
with any deviation from linearity would not result in a
significant error for IMRT analysis.
For scanners that operate in transmission mode (Canon
and V370), the graphs were best fit and did not indicate the
amount of system noise, so the useable range of the V370
was from 0 to approximately 20 Gy and the Canon from 0
to approximately 3 Gy. However, since the most common
fraction dose in radiotherapy is 2 Gy, these scanners can
still be considered useful.
It was noted that the system response of the Red channel
for those scanners that operated in transmission mode
(V800 and V700) was generally greater than for the Green
Fig. 2 Calibration graph for V700 Green channel to 300 cGy
comparing processed measured data versus Power Fit versus Linear
reference
Fig. 3 Data at 1 h post exposure, prepared with calibration constant for 1 h
Australas Phys Eng Sci Med (2016) 39:533–545 537
123
Fig. 4 Data at 16 h post exposure. Note approximately 8 % further development of film
Fig. 5 Data at 16 h analysed using relative mode. (100 % of points pass)
538 Australas Phys Eng Sci Med (2016) 39:533–545
123
channel. However, a pure system response is not as
important as the resulting signal to noise ratio. This system
has 2 major components, the film and the scanner, both of
which are sources of noise. The noise associated with the
scanner can be reduced by increasing the sample time. If
the resulting increase in scan time is not significant, to
achieve the same signal to noise ratio for the Green channel
as for the Red channel, then the greater sensitivity of the
Red channel should not be a major consideration in the
overall choice.
The recommendation to use the Green channel was
based on the overall properties; especially the property that
it was approximately linear with dose. This does not pre-
clude the use of the other channel data, which for certain
applications may be suitable, it is a recommendation for
general use that enables a simple method that can be easily
understood and implemented.
To address the issue of system noise and scanner stability
it was determined to scan at a significantly higher spatial
resolution than required. With the Epson scanners this
increases the scan time, which in turn stabilises the response
of the scanner and as the high resolution pixels are down
filtered to the required resolution (1 mm/pixel for IMRT
QA), eliminates a significant portion of the system noise.
To address the issue of noise introduced by dividing by
unexposed scan, a Median filter effectively removed noise
without affecting the spatial and amplitude resolution very
much. A median filter (Gonzalez and Woods [14], Chap. 3)
is a type of Order-statistic (nonlinear) filter. Median filters
provide excellent noise reduction with considerably less
blurring than linear smoothing filters and are particularly
effective for impulse noise.
Explanation of calibration values
Since the data had already been multiplied by a constant
such that a measured value of 200 equates to 200 cGy, then
in the calibration window a value of 200 was entered for
the measured and response. This was also the case for the
zero value, where from derivation a measured value of zero
equates to a zero dose. However, the response function for
the systems was only approximately linear which caused a
discrepancy at values other than 0 and 200. The discrep-
ancy at 100 cGy was approximately 5 %, so a measured
value of 105 equated to 100 cGy. When a calibration
function of the form y = Ax ^ B (Power) was applied, it
gave a good agreement in the range to approximately
300 cGy. This function has 2 degrees of freedom (A, B),
and as such is fully specified by the 3 data points.
Note again that there may be a small difference in the
amplitude scale when this data is compared to the planned
data in the IMRT comparison package, but it can be cor-
rected by selecting the Relative comparison mode in the
IMRT analysis package. This is valid because the cali-
brated data is effectively directly proportional to dose.
Referring to Table 1, the data are for the range 0–3 Gy.
For the range beyond 3–50 Gy it is not necessary to apply
the correction to meet the IMRT criteria discussed. As
discussed the function is only approximately linear,
therefore for a different target dose, for example 15 Gy, the
proportional constant for the V800 is 675 and for the V700
is 750. This results in approximately a 4 % absolute error
for values near 2 Gy, however as the VanDyk criteria [15]
is generally applied for IMRT, then the global error is
approximately 0.5 %. This assumes that if the target dose is
15 Gy, then the maximum dose will be close to 15 Gy.
Examples of clinical QA use
An example set of results for a Prostate IMRT treatment is
presented below Figs. 3, 4 and 5.
A general criterion for assessing IMRT is by the Gamma
method [16] using 3 % and 3 mm applied as global.
Applied as global, which means calculating the percentage
difference relative to the maximum value of all the points
assessed [15]. However, since film is considered to be a
high spatial resolution device it is common to use 4 % and
2 mm. All of the scanner systems returned accept-
able passing results, achieving[99.5 % of points passing
the above criteria for a clinical IMRT treatment. Note 3 %
3 mm and 4 % 2 mm are the assessment criteria commonly
used at NCCI for IMRT analysis.
Table 1 Tabulated values of a pixel exposed to 200 cGy, the resulting proportional constant and the correction required at 100 cGy for each of
the scanners in the range 0–300 cGy
Processed value of pixel
exposed to 200 cGy
Proportional
constant
Correction at
100 cGy
Epson V800 0.308 650 103.0
Epson V700 0.282 710 105.0
Epson V370 0.760 263 105.0
Canon 0.714 280 107.0
Gafchromic EBT3 film
Australas Phys Eng Sci Med (2016) 39:533–545 539
123
This set of results was obtained using the Epson V700,
but all the other evaluated scanners achieved[99.5 % %
of points passing for this case.
The initial scan was acquired at 1 h post exposure. A
subsequent scan for each scanner was acquired at 16 h post
exposure, and it was found that the derived function had
increased by approximately 8 %. This result was then
assessed using the Relative mode of application SNC
Patient, which allows normalisation between the TPS cal-
culated and the measured dose. The second example is the
subsequent scan assessed in Relative mode where once
again, all the assessments returned [99.5 % of points
passing.
Conclusion
Various scanners were used and an appropriate calibration
method that encompasses these scanners has been outlined.
The results are generally accurate to the tolerance required
for IMRT analysis, for example, Gamma criteria of 3 %
3 mm or 4 % 2 mm. This means the film/scanning system
errors were much less than this criteria and as such, a valid
analysis was achieved for treatment delivery.
The recommendation would be to use the Epson V800
scanner based on the characteristics of the LED based light
source. The Green channel appears to be well matched to
Gafchromic film such that a near linear response was
achieved from the derived function from 0 to 50 Gy. This
range covers almost all realistic treatment fractions,
including those associated with SBRT and SABR tech-
niques. Furthermore it appears to be more stable than the
Cold Cathode light source of the V700, both from scan to
scan and in the uniformity of the response across the
scanner. The V800 scanner is not that expensive, being
approximately the same cost as a standard box of Gaf-
chromic film.
However, if no specialised scanner is available, the
method showed that acceptable results in the range 0–3 Gy
were achievable with general document scanners such as
the Canon multi-function office unit evaluated, and which
are ubiquitous in modern departments. Furthermore, most
of these units can scan in A3 size, which although not
required for Patient QA dosimetry, is useful for some
machine dosimetry where measurements are required over
a large field.
The Epson V37 series was interesting because it also
uses an LED light source and the Green channel had a well
behaved and stable response over 20 Gy. However the
linearity of this response was only reasonable to 3 Gy.
Higher exposures would require a more complex calibra-
tion function. Because the V700 and V800 scanners which
operate in transmission mode and are superior, were not
that expensive, no further investigations were carried out
on the V37 series.
Acknowledgments Image processing package ImageJ available
from: National Institutes of Health, Bethesda, MA, USA Web site:
http://imagej.nih.gov/ij. Version used was ImageJ 1.48v Note some
functions may not be available in earlier versions and the authors have
no control over future versions of ImageJ. Gafchromic refers to a
range of Radiochromic film products supplied by: Ashland Speciality
Ingredients, International Speciality Products, 1361 Alps Road
Wayne • NJ, USA. Primarily Gafchromic EBT3 from Lot# 07281402
was used for this study. The Epson range of scanners was supplied by:
Seiko Epson Corp.Nagano, Japan. The IMRT evaluation program
used was SNC Patient v6.5 supplied by: Sun Nuclear Corporation,
Melbourne, FA, USA.
Appendix 1
Use of a scanner with 8 bit dynamic range
in reflective mode
The basic problem of using 8 bit resolution is that it only
allows 256 values to cover the range of interest. If, for
example, we were interested in a dose range from 0 to
250 cGy, this could possibly allow 1 cGy per division as a
best case. However, the Canon office scanner returned a
pixel value (in the Green channel) of 0 cGy as 136 and for
250 cGy the pixel value was 72. Therefore there were only
64 divisions to cover 250 cGy, which resulted in 4 cGy per
division. Furthermore, because the response was non-linear
the step per division will be greater at parts of the range.
Note that this was the raw scanner values, not the derived
function. This resulted in an amplitude resolution of
approximately 2 % of the range. A system resolution of
better than 1 % is generally required for IMRT QA, in
order to discern dosimetric errors of the order of 3 %.
Acceptable amplitude resolution can be achieved by
scanning at a high spatial resolution and then down filtering
to the required spatial resolution (generally 1 mm/pixel),
while enhancing the amplitude resolution in the process.
Note that aliasing was addressed as the light scattering
properties of Gafchromic film led to a distribution of values
for any measurement. This distribution acted effectively as
an anti-aliasing filter which, for the Canon printer, worked
well in the range from 0 to 3 Gy.
Appendix 2
Using an Epson scanner and the Epson Scan
software application
The specific models used were the Epson V800, V700, and
V370.
540 Australas Phys Eng Sci Med (2016) 39:533–545
123
It was assumed that the scanner had been set up
according to the manufacturer’s recommendations and
connected to a PC (Windows 7), and the Epson Scan
software had been loaded.
A 25 9 20 cm film area guide or template should be
prepared for the V800 and V700 scanners such that the
piece of film may be placed accurately in the active area of
the scanner. An area guide was supplied with the V800 and
V700 scanners, but a simple right angle template was found
to be more practical.
For reflective mode, the scan area can extend to the
scanner bezel. The film can be accurately located against
the edge of the bezel so a separate area guide is not
required.
Activate the Epson Scan software application and ensure
that the acquisition parameters are as specified, that all the
colour corrections are switched off, then acquire RGB 48
bit, and save as file type.TIF.
Acquire the full area of the film (254 9 203 mm) and
specify a high spatial resolution such as 508 dpi (20 dpmm).
Save the pre-exposure ‘‘Blank’’ and the post-exposure
‘‘Exposed’’ scans in a suitable working directory. Since
these files can be large, it is recommended to save them on
the local hard drive of the scanner workstation.
Using the Canon office scanner
The Canon scanner works in reflective mode and the scan
area extends to the scanner bezel. The edge of the bezel can
be used to accurately locate the film so a separate area
guide is not required. However, ensure that the film is re-
positioned accurately between the Blank scan and Exposed
scan. Since these are general office scanners, the scanner
should be replicated between scans as far as possible, so be
aware that other users may leave unwanted marks on the
scanner.
To acquire a scan:
Select maximum spatial resolution (for the Canon this
was 600 dpi)
Select Colour
Select file type JPG (note several other file types are
acceptable to ImageJ, JPG is for Canon)
Select save file option. On the Canon the file is sent to
the users email address.
The Canon does not save the file in RGB format, so to
prepare the file for processing, open in ImageJ.
Use the Rectangle select tool to mark the extent of the
film.
Use Image ? Crop. This extracts the specified area.
Use Image ? Type ? RGB stack. This converts the
image to Type RGB
Save as.TIF to the working directory
This should be done for both the Blank and Exposed.
Note regarding JPG file format used by the Canon
scanner. JPG is the image format used by the Canon
multi-function scanner. Taken in context that software is
part of the system, the use of JPG by Canon is a com-
ponent of the resolution that can be achieved by the
scanner. The operator has no control over the compres-
sion level used. It could be assumed that it is optimised
by Canon, however the end point is the resolution that is
achieved considering the unit as a system including the
software.
Process data using application ImageJ
Data processing in ImageJ is done using ImageJ Macros.
These are scripts which list the commands. They also form
a summary documentation of the required steps. They are
listed below.
To run a macro in ImageJ, use:
File ? Open and select the macro (it should be saved
with extension.ijm)
The Macro window should open, then from the Menu
bar on the Macro window:
Select Macros ? Run Macro
It is assumed the scanned image files (Blank & Exposed)
are in the working directory.
Using ImageJ:
Open image file (RGB) of the Pre-Exposure image
(Blank).
Run macro: IJmacro_PreProcess_BLANK.ijm
Open image file (RGB) of the Post-Exposure image
(Exposed).
Run macro: IJmacro_PreProcess_EXPOSED.ijm
The pre-processed but uncalibrated image of the exposed
film should be open at this point.
If there are faults in the image such as pin holes from the
phantom, these can be edited. To remove pin holes, draw a
small rectangle around the pin hole, and then apply theMedian
filter with a radius 20. When complete, proceed to next step.
Australas Phys Eng Sci Med (2016) 39:533–545 541
123
Run macro: IJmacro_Calibrate_V800_EBT3_LowRange.
ijm
The calibrated data is now saved as a Text Image in file:
Dose_Cal_Text_254x203.txt
This file can be converted to a format that can be read in
conjunction with software application SNC Patient using
the following steps:
Convert the file to a.csv file using MS Excel. Open the
Text Image file using Excel, specify Tab delimited data.
Then save file as type.csv (comma separated variable).
Pre-pend a header, such that the application SNC Patient
recognises the data as type XiO.
Use Batch file: Add_Header_to_csv_file.bat. This batch
file can be run from a CMD prompt with the name of the
file to be converted as 1st parameter. See below. Or right
click the file to be converted and select Open with, then
select batch file.
C:\!Data\254data[Add_Header_to_csv_file.bat
Dose_Cal_Text_254x203.csv
This batch file is listed below. The converted file can
now be read into the application SNC Patient. The
data is now in units of Dose (cGy). The file name is:
Dose_Cal_Text_254x203.csvX
Macro: IJmacro_PreProcess_BLANK.ijm
// This ImageJ macro converts a RGB to single channel (G) 32b 1016x812// The converted image is saved in same dir with name BLANK.tif//print("IJmacro_Preprocess_BLANK v1 24june2015")print(" ")//// Note it is Required that the Blank RGB file is open and is the only image open in ImageJ.//Fname = getTitle;print("Filename = "+Fname);//WDir = getDirectory("image");print("Working directory: "+WDir);//run("Stack to Images");selectWindow("Red");close();selectWindow("Blue");close();//selectWindow("Green");run("32-bit");run("Mean...", "radius=2");run("Rotate 90 Degrees Left");run("Size...", "width=1016 height=812 average interpolation=Bilinear");run("Grays");saveAs("Tiff", WDir+"BLANK.tif");close();print ("Macro complete");// // End
542 Australas Phys Eng Sci Med (2016) 39:533–545
123
Macro: IJmacro_PreProcess_EXPOSED.ijm
// This ImageJ macro converts a RGB to single channel (G) 32b 1016x812// The converted image is saved in same dir with name EXPOSED.tif//// The first stage of process is then run, using the stored Blank data.// print("IJmacro_PreProcess_EXPOSED v1 24june2015")print(" ")//// Note it is Required that the Exposed RGB file is open and is the only image open in ImageJ.// it is required that the processed Blank is stored in filename: BLANK.tif//Fname = getTitle;print("Filename = "+Fname);//WDir = getDirectory("image");print("Working directory: "+WDir);//run("Stack to Images");close("Red");close("Blue");//selectWindow("Green");run("32-bit");run("Mean...", "radius=2");run("Rotate 90 Degrees Left");run("Size...", "width=1016 height=812 average interpolation=Bilinear");run("Grays");saveAs("Tiff", WDir+"EXPOSED.tif");close();print ("PreProcess Exposed RGB data complete");//open(WDir+"\\EXPOSED.tif");open(WDir+"\\BLANK.tif");//imageCalculator("Divide create 32-bit", "BLANK.tif" , "EXPOSED.tif" );//close("BLANK.tif");close("EXPOSED.tif");////selectWindow("Result of BLANK.tif");run("Subtract...", "value=1");run("Median...", "radius=2");run("Min...", "value=0");run("Multiply...", "value=1000");resetMinAndMax();saveAs("Tiff", WDir+"RPDx1000_32b_1016x812.tif");print ("RPD Raw Processed Data written to file: RPDx1000_32b_1016x812.tif");run("Size...", "width=254 height=203 average interpolation=Bilinear");
Australas Phys Eng Sci Med (2016) 39:533–545 543
123
saveAs("Text Image", WDir+"RPDx1000_Text_254x203.txt");close();//open(WDir+"RPDx1000_32b_1016x812.tif");//// Leave RPD file open, ready to run appropriate calibration macro.//print ("Macro complete");// // End
Macro: IJmacro_Calibrate_V800_EBT3_LowRange.ijm
// This ImageJ macro applies a Calibration to a 32b 1016x812 image// The converted image is saved in same dir as both a 32b Tif and as a 254x203 Text Image//print("IJmacro_Calibrate_V800_EBT3_LowRange v1 24june2015")print(" ")//// Note that the 32b 1016x812 image MUST be open and is the only image open in ImageJ.//Fname = getTitle;print("Filename = "+Fname);//WDir = getDirectory("image");print("Working directory: "+WDir);////// Enter Rm200 value - EBT3 with V800 -> approx 650// Note data is x1000// Note set max value is dependent on range, expected max, change as appropriate//run("Multiply...", "value=0.65");run("Min...", "value=0");run("Max...", "value=400");//// Prepare to run Calibrate - create a 16b window// Note do not convert direct to 32b as it will re-scale the dataselectWindow(getTitle);run("Copy");run("16-bit");run("Paste");resetMinAndMax();//
run("Calibrate...", "function=Power unit=[Gray Value] text1=[0.0 103.0 200.0 ] text2=[0.0 100.0 200.0 ]");//// Covert back to 32brun("32-bit");saveAs("Tiff", WDir+"Dose_Cal_32b_1016x812.tif");//run("Size...", "width=254 height=203 average interpolation=Bilinear");saveAs("Text Image", WDir+"Dose_Cal_Text_254x203.txt");close();////print("Calibrated Data saved to: Dose_Cal_32b_1016x812.tif & Dose_Cal_Text_254x203.txt ");print("Macro Completed ");exit// End
544 Australas Phys Eng Sci Med (2016) 39:533–545
123
References
1. GafchromicVR EBT (2010) Self-developing film for radiother-
apy dosimetry. ISP White Paper
2. Lewis D, Chan MF (2015) Correcting lateral response artifacts
from flatbed scanners for radiochromic film dosimetry. Med Phys
42(1):416–429
3. Williams MJ, Metcalfe PE (2011) Radiochromic film dosimetry
and its applications in radiotherapy. University of Wollongong,
Wollongong
4. Butson MJ, Cheung T, Yu PKN (2006) Scanning orientation
effects on Gafchromic EBT film dosimetry. Australas Phys Eng
Sci Med 29(3):281–284
5. Alnawaf H, Peter KN, Butson M (2012) Comparison of Epson
scanner quality for radiochromic film evaluation. J Appl Clin
Med Phys 13(5):3957
6. Micke A, Lewis DF, Yu X (2011) Multichannel film dosimetry
with nonuniformity correction. Med Phys 38(5):2523–2534
7. Kairn T, Aland T, Kenny J (2010) Local heterogeneities in early
batches of EBT2 film: a suggested solution. Phys Med Biol
55(15):L37
8. Hu Y, Wang Y, Fogarty G, Liu G (2013) Developing a novel
method to analyse Gafchromic EBT2 films in intensity modulated
radiation therapy quality assurance. Australas Phys Eng Sci Med
36(4):487–494
9. Borca VC, Pasquino M, Russo G, Grosso P, Cante D, Sciacero P,
Tofani S (2013) Dosimetric characterization and use of GAF-
CHROMIC EBT3 film for IMRT dose verification. J Appl Clin
Med Phy 14(2):4111
10. Hardcastle N, Basavatia A, Bayliss A, Tome WA (2011) High
dose per fraction dosimetry of small fields with Gafchromic
EBT2 film. Med Phys 38(7):4081–4085
11. Mendez I, Hartman V, Hudej R, Strojnik A, Casar B (2013)
Gafchromic EBT2 film dosimetry in reflection mode with a novel
plan-based calibration method. Med Phys 40(1):011720
12. Rasband WS (2012) ImageJ: image processing and analysis in
Java. Astrophys Source Code Libr 1:06013
13. Abramoff MD, Magalhaes PJ, Ram SJ (2004) Image processing
with ImageJ. Biophotonics Int 11(7):36–42
14. Gonzalez RC, Woods RE (2008) Digital image processing, 3rd
edn. Pearson Prentice Hall, Upper Saddle River
15. Low DA, Dempsey JF (2003) Evaluation of the gamma dose
distribution comparison method. Med Phys 30(9):2455–2464
16. Van Dyk J, Barnett RB, Cygler JE, Shragge PC (1993) Com-
missioning and quality assurance of treatment planning comput-
ers. Int J Radiat Oncol Biol Phys 26(2):261–273
Batch File to Pre-pend header: Add_Header_to_csv_file.bat
ECHO OFF
REM Add_Header_to_csv_file.bat v1 24June2015 REM Batch file to Pre-pend a XiO format header to a dose plane file for SNC Patient. REM Data should be in text format as a .csv file REM Normally this is prepared from an ImageJ text image (.txt), which is converted to .csv using MS Excel REM Run by Right click on file to be converted, and then select Open With ->Add_Header_to_csv_file.bat REM Can also be run from the command line with the name of file to be converted as 1st parameter.
ECHO %~1
CMD /C " ECHO X000111,,,,, > "%~1X" " CMD /C " ECHO DateTime,D18/12/2014,DocNum 1234,D11223344,, >> "%~1X" " CMD /C " ECHO PatientID,P111111,,,, >> "%~1X" " CMD /C " ECHO PlaneDesc,T: 0.00 cm,,,, >> "%~1X" " CMD /C " ECHO DoseUnits, cGy,-100, -1.0 cGy, Abs, >> "%~1X" " CMD /C " ECHO CompfType,-1,,,, >> "%~1X" " CMD /C " ECHO FieldSizeDefDist,-1,,,, >> "%~1X" " CMD /C " ECHO CollWidLenQAplane,-1,-1,,, >> "%~1X" " CMD /C " ECHO OutputWidLenQAplane,-1,-1,,, >> "%~1X" " CMD /C " ECHO QAssd,-1,,,, >> "%~1X" " CMD /C " ECHO QAdepth,-1,,,, >> "%~1X" " CMD /C " ECHO QAedens,-1,,,, >> "%~1X" " CMD /C " ECHO Upperleft,-125,110,,, >> "%~1X" " CMD /C " ECHO CalcGridResmm (x,y,z),1,1,1 >> "%~1X" " CMD /C " ECHO DosePtsxy,254,203,,, >> "%~1X" " CMD /C " ECHO DoseResmm,1,,,, >> "%~1X" "
CMD /C " TYPE "%~1" >> "%~1X" "
REM PAUSE EXIT
Australas Phys Eng Sci Med (2016) 39:533–545 545
123