Computer Methods and Programs in Biomedicine 212 (2021) 106457

Contents lists available at ScienceDirect

Computer Methods and Programs in Biomedicine

journal homepage: www.elsevier.com/locate/cmpb

Profeel—An open source dosimetry data visualization and analysis

software
Tomppa Pakarinen a,b,∗, Jarkko Ojala a
a
Department of Medical Physics and Department of Oncology, Tampere University Hospital, Tampere, Finland
b
Faculty of Medicine and Health Technology, Tampere University, Tampere, Finland

a r t i c l e i n f o a b s t r a c t

Article history: Background and objectives: This article presents Profeel, a Matlab (MathWorks Inc., MA) based open
Received 30 April 2021 source dosimetry data visualization and analysis software. Profeel aims to support quality assurance,
Accepted 11 September 2021
dosimetry and research in the field of radiotherapy by providing an environment to visualize, process and
analyse measured and simulated dosimetry data from several data sources used in radiotherapy practice
Keywords: and research.
Dosimetry Methods: The processing and analysis tools are based on routinely used dosimetry analysis methods, such
Matlab as gamma analysis, different data normalizations and data filtering. Additionally the Profeel performs an
Radiotherapy automatic 1 dimensional profile and percentage depth dose analysis in accordance with International
Quality assurance
Electrotechnical Commission definitions. All data can be operated by user created custom functions and
lower dimensionality data can be extracted from volume doses and dose planes.
Results: Profeel supports data import in all 3 dimensions and offers an intuitive user interface to perform
data visualization, processing and analysis between simulated and measured data. Profeel and its source
code are distributed free of charge under the General Public Licence (GPL).
Conclusions: Profeel has shown to be an agile tool for fulfilling various needs of several researchers and
since Profeel is under constant development and is an open source project, community needs, issues and
bug reports are taken into account in the development.
© 2021 The Author(s). Published by Elsevier B.V.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)

1. Introduction if left uncorrected. These include volume averaging due to finite

size of the active volume, energy and directional dependence etc.
In radiotherapy (RT), the radiation dose determination is essen- Also, the measured results always represent a single sample of the
tial in assuring an accurate treatment delivery, maximizing the po- output of the device producing the radiation beam to be measured.
tential health benefit and minimizing the possibility for a poten- In dose calculation the challenges are usually related to the ap-
tial adverse health effect to the patient. Dose determination can proximations in the underlying dose calculation algorithm simu-
be done through dosimetric measurements (dosimetry) and dose lating the radiation transport in matter.
calculations. In modern RT, three dimensional (3D) dose distribu- Thus, to achieve high quality data and high level of confidence
tions are to be determined by default, from which two dimensional in dose determination, it is essential to combine data from vari-
(2D) and one dimensional (1D) data sets are extracted for further ous sources, taking into account their related uncertainties. This re-
analysis and use. Also, in measurements, often 1D and/or 2D data quires the comparison of data, but often the software for data visu-
can be directly acquired. alization and analysis support only the data format from the mea-
No matter which method is utilized in dose determination, surement device manufacturer itself and even if general purpose
there are always various sources of uncertainty involved. In data formats are supported, lots of pre-processing is required prior
dosimetry, the equipment, especially the detector, introduces in- to successful import. Therefore, in this work, we have developed
trinsic and extrinsic factors that add to the combined uncertainty, a software, that supports measurement data formats from several
dosimetry device manufacturers, general dose distribution data for-
mat used in clinical RT treatment planning systems (TPS) (DICOM-
∗
Corresponding author at: Postal: Tampereen yliopistollinen sairaala Säde-
RT) and Monte Carlo dose calculation data format (EGSnrc/3ddose),
hoitoyksikkö PL 20 0 0 33521 Tampere, Finland. which are often used for independent validation of dose distribu-
E-mail address: Tomppa.pakarinen@tuni.fi (T. Pakarinen). tions.

https://doi.org/10.1016/j.cmpb.2021.106457
0169-2607/© 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)
T. Pakarinen and J. Ojala Computer Methods and Programs in Biomedicine 212 (2021) 106457

There are few published works considering open source pro- ‘analytical’ refers to solving the line or plane cross sections ana-
grams and tools for radiotherapeutic dosimetry data analysis. Pro- lytically in cases where the initial linear interpolation sampling is
grams, such as Computational Environment for Radiological re- not adequate, e.g. when lower dimension extraction is requested
search (CERR) [1] SlicerRT radiation therapy toolkit for 3D slicer from a point that does not exist in the data structure. The user is
[2] and XSTING [3] are developed tools for radiotherapeutic re- prompted to wait during data import, since large dataset import
search purposes, but do not support measured data import and times may be long (minutes). For .3ddose data type, the measure-
thus, computed and measured data comparison. Other published ment uncertainty (‘error’ in Profeel) dataset can optionally be ne-
dosimetry and dose calculation software often require custom glected, thus halving the import time. Otherwise the uncertainty
measurement setups [4], are intended on fields outside of radio- data will be imported as default.
therapy [5] or focus on radiotherapy treatment planning and op- Data- and metadata structures are added to the main struc-
timization [6,7], and rarely supply the source code. Unpublished ture during lower dimension extraction, similar to the structure
work such as 3ddose-tools (.3ddose) [8] and web-based VICTORIA transformations during the data import. If the user-requested plane
(.3ddose, .dcm, .egsphant) [9] allow the user to import, visualize does not exist in 3D stack, Profeel interpolates the plane from the
and plot computed data but unfortunately do not support mea- requested position. For analogous cases in 1D, Profeel finds the
sured data formats. At the time and until today, to the best of our nearest existing points from the linearly interpolated data sub-
knowledge, there are no existing open-source programs to visual- structure.
ize, process and analyze the required source data (see section 3.1)
from a dosimetric standpoint. 2.3. Interactive profile

2. Methods Quick profile is an interactive tool in Profeel, which allows the

user to display and adjust 1D profiles from 2D data. In Profeel, 2D
Profeel v. 1.17 is programmed completely with Matlab (R2020a data structure is saved as m x n matrix M, where m and n corre-
v. 9.8), utilising graphical user interface development environment spond to the visualized 2D data dimensions, i.e. either the origi-
GUIDE creation library, Image ProcessingTM , Signal ProcessingTM nal or lower resolution data. A line L, drawn within M consist of
and Parallel ComputingTM Toolboxes. Matlab is used extensively a starting point p1 and and an endpoint p2 , which confine a sub-
in radiotherapy research, with a variety of build-in Toolboxes and matrix m, where m⊆M, and where the diagonal diag(m ) = L =
functions for DICOM and other technical extensions. Profeel com- p1 − p2 . Depending on the line orientation, m is reflected so that
bines dosimetry data visualization with processing and analytics the former is always true, i.e. final line profile is the submatrix di-
tools in an intuitive and easy to use user interface (UI). Profeel has agonal.
a number of data processing and analytic UI tools, and the details
of the computationally most relevant methods are described in de- 2.4. Gamma analysis
tail below.
Profeel includes gamma analysis for 1D and 2D data. The
2.1. Processing during import and data extraction gamma index computation is performed for each data point by
minimizing the squared sum of the gamma components (DTA, DD)
First, all imported data is combined into a structure set with relative to their respective gamma criteria. Gamma indices for are
predefined fields. The main data structure consists of sub struc- calculated as presented in Eq. (1)
tures containing the raw data, interpolated data, data normaliza-
tions as matrices with corresponding position vectors. Addition-
γ (rm ) = min {(rm , rc )}∀{rc } (1)
ally, result fields, reference to parent structure and metadata, are where rm and rc refer to the measured and calculated spatial loca-
included in the structure. Visualization of the imported data, es- tions and where
pecially for large 3D datasets, requires processing time. To reduce 
the wait time during visualization, a lower resolution dataset is |rc − rm |2 |Dc (rc ) − Dm (rm )|2
(rm , rc ) = + (2)
automatically computed and added to the structure, if the origi- d m
2 D2m
nal resolution exceeds a predefined hard-coded value of 120 data
points per axis, which was considered as a good compromise be- In Eq. (2), |Dc (rc ) − Dm (rm )|2 is the squared difference between
tween visual details and computing performance. The low reso- measured and calculated doses (target and reference), and dm
lution dataset is used only to speed up visualization, and as the mand Dm are the DTA and DD criteria, respectively. [10] Profeel
default option for gamma index, distance to agreement (DTA) and uses an open-source gamma computation function to compute the
dose difference (DD) computation, when the low resolution option final gamma indices [11]. Additionally, DTA and DD can be com-
is activated. puted with the gamma indices or DTA and DD computation can be
performed alone. DTA and DD are computed independently from
2.2. Interpolation methods the gamma function.

Imported 1D and 2D data are automatically interpolated with 2.4.1. Data preparation for gamma analysis
linear interpolation to 10 times the original resolution, i.e. values Profeel prepares the data for gamma computation by automat-
are added linearly between successive data points, preserving the ically matching the spatial dimensions between the two datasets
original points. Interpolated data is then added to the data struc- by truncating the largest axes. Hence, analysed input data does not
ture. 3D data is not interpolated during import, apart from lower have to have the same spatial dimensions. Both datasets are also
resolution interpolation. The interpolated data is used as the de- interpolated to the same resolution, if their physical sampling di-
fault data source for profile and PDD parameters, and for ‘brute mensions disagree, so that gamma, DTA and DD can be computed
force’ method computations. Additionally, interpolated data is used for each data point. The default sampling resolution is equal to the
as the source data in ‘analytical method’ computation for gamma dataset with lower resolution, but the used resolution can be mod-
parameters and lower dimension data extraction. Here the ‘brute ified by the user.
force’ method refers to additional interpolation, finding the clos- The gamma analysis results are sensitive to the statistical fluc-
est values in a numeric manner through linear interpolation, and tuations present in MC dose distributions, as described by Graves

2
T. Pakarinen and J. Ojala Computer Methods and Programs in Biomedicine 212 (2021) 106457

Fig. 1. 2D DTA computation error due to finite sampling.

et al. [12] Therefore the user of Profeel has an option to apply Table 1
Supported data types in Profeel.
the smoothing prior to gamma analysis and to choose whether
to use MC dose distribution as reference or as evaluation dataset. Data type Extension
Additionally, the user can use custom functions to operate both PTW beamscan exports .mcc
input datasets along with several different normalization options IBA Omnipro exports ASCII, .OmniIMRT, .opg
(section 3.1). PTW software exports ASCII
EGSnrc output files .3ddose.dcm
2.5. Error determination due to finite grid sampling DICOM RT RD dose files
Matlab structures compatible with Profeel .mat
(previous sessions)
Finite sampling introduces error to the computed DTAs, which
further transfers to the gamma-indices. The user is informed about Table 2
the error in the DTA result legend. The origin of the error is pre- Normalizations in profeel.
sented in Fig. 1.
Normalization type Applicable data
DTAs are computed within a moving neighbourhood on a sur-
face S⊂Z2 : for which, function f: S → R and reference dose point No normalization 1D, 2D 3D
Normalization to maximum 1D, 2D 3D
c ∈ R. Let’s define the neighbourhood H s.t.
Normalization to central axis (CAX) 1D
H := {(r1 , r2 ) ∈ S|r1 − r2  ≤ ( f (r1 ) − c ) · ( f (r2 ) − c ) < 0}, (3) Distance percentMean value from range 1D1D
Normalization to value 1D, 2D 3D
where maximum
√ error within the neighborhood becomes ξ =
r2 − r1  = 2 · resolutionx,y , for spatially isotropic square grids. As
demonstrated in Fig. 1, the spatial order of D2 and D3 from the ref-
All data with the same dimensionality can be visualized and
erence dose is known, thus the actual DTA must lie within ξ from
processed simultaneously as presented in Fig. 2.
D2. This only accounts for error due to finite grid sampling in DTA
Profeel offers several data processing and analysis tools with
algorithm.
varying flexibility, and data and result exportation in .mat and .csv
3. Results formats for further processing and analysis. Data saved in .mat
form is equivalent to saving the work session and can be imported
Profeel open-source project was originally created as a support- to Profeel as such.
ive tool for intuitive dosimetry data comparison and analytics be-
tween different data sources used in radiotherapy quality assur- 3.2. User interface
ance (QA). Subsequently project development has been directed to-
wards research purposes. The project was initialized in 2020 at the The Profeel graphical user interface (GUI) is built in the Mat-
Department of Oncology at Tampere University Hospital (Tays). lab (MathWorks Inc., MA) R2020a graphical user interface devel-
opment environment (GUIDE). Profeel GUI is presented in Fig. 3,
3.1. Supported data types for which, the GUI object groupings are presented in Table 3.
The UI figure, visualizing the data is referred as the ‘main fig-
Profeel (v. 1.17) supports several different data types, includ- ure’. After data import, the data chosen from group b) is presented
ing commonly used dosimetry measurement equipment outputs in in the main figure. In the example in Fig. 3, a 3D DICOM RT dose
radiotherapy QA, and computed or simulated output file formats dataset is chosen. 3D data is saved during import to a Matlab
listed in Table 1 structure as 2D stacks, which can be navigated through with the
There are no limitations for the number of imported data other mouse scroll wheel. 2D and 3D data are always presented as 2D
than set by the amount of computer memory. All datatypes are slices/planes in Profeel.
reconstructed to common structural form (.mat), containing raw,
processed, result, parent reference and metadata fields. 3.3. Data processing tools
Lower dimension data can be extracted from higher dimen-
sions, i.e. 2D and 1D from 3D and 1D from 2D. Available data nor- Data processing tools are mainly located in GUI grouping c),
malizations are presented in Table 2. under the ‘Main data processing panel’. The panel includes sub-

3
T. Pakarinen and J. Ojala Computer Methods and Programs in Biomedicine 212 (2021) 106457

Fig. 2. Simultaneous visualization of measured and simulated dose profiles.

Fig. 3. Profeel user interface.

Table 3
Profeel GUI groupings in Fig. 3.

GUI group name Description

a) Data import panel Data import button and 1D CAX correction during import
b) Imported data listbox Contains all imported data. Data is activated by mouse click.
c) Main data processing panel Orientation, scaling, data filtering and custom operations
d) Profile and PDD parameters(See chapter 3.4) Shows automatically computed 1D profile and PDD parameters
e) Data normalization panel Different normalizations (see Table 2)
f) Gamma, DTA and DD panel Gamma analysis tools
g) Data exportation panel .csv and .mat data export
h) Tool ribbon Additional and future tools

4
T. Pakarinen and J. Ojala Computer Methods and Programs in Biomedicine 212 (2021) 106457

Fig. 4. Data processing GUI panel.

sections consisting of 1D and 2D/3D standard tools, such as data

mirroring, inversion, rotation, shifting operator and low resolution
visualization. Under the drop down menu in the upper right corner
of Fig. 3 are additional processing tools, i.e. Savitzky-Golay smooth-
ing filter for 1D and 2D data [13] and a data scaling tool (1D, 2D
and 3D) as presented in Fig. 4.
Under the dropdown menu is located the ‘Add operators’ GUI
button, which allows the user to use custom processing functions Fig. 5. Lower dimension data extraction button locations and the input prompt
via Matlab syntax. The creation of custom processing files (ASCII) pop-up for lower dimension data extraction.
is described in detail in the Profeel Git repository’s documentation
[14]. Custom processing functions are applicable for all imported
data types in all 3 dimensions. All front end processing operations
are implemented directly on chosen data and visualized in the GUI 3.4. Data analysis tools
main figure.
The data normalization drop down menu is placed under the 3.4.1. Profile and PDD parameters
normalization panel (GUI group e), under the import button. Data Data analysis options and tools are distributed within GUI
normalization options are described in section 3.1 and more de- groupings d), f), h), and b) in Fig. 3, where the group d) contains
tails, if required, are explained in the Profeel documentation [14]. the profile and PDD parameters. Profeel classifies the 1D curve
Data extraction refers to the extraction of lower dimension data types automatically during import and lower dimension extrac-
from higher dimensions as described in section 2. Extraction op- tion. The automatic curve type recognition is based on the curve
tions appear inside group b), under the listbox object after the symmetry and field size. The user may also re-assign the curve
given data is activated. After choosing the extracted dimensional- type under the tool ribbon (group h), which forces the program
ity, the user inputs the extracted axis and the range to a pop-up to compute the assigned parameters. Profile and PDD parameters
prompt as show in Fig. 5. are computed based on IEC standard 60976 [15]. The profile and
If the user wants to inspect line profiles from the 2D data be- PDD parameters are listed in Table 4 and 5, respectively.
fore 1D extraction, an interactive quick profile tool can be used, Flattened region for profiles is defined only for field sizes larger
presented in Fig. 6. or equal to 5 cm in the IEC Standard 60976. Currently Profeel has
Quick profile is chosen from the ‘Quick profile’ button included no explicit definition for field sizes below 5 cm, thus the flattened
in the GUI group b). The button is visible only for applicable data region is computed as for 5 cm≤Field size≤10 cm. [15]
(2D). If the user has performed gamma or DTA and DD analysis for If any of the parameter computation fails or profile field size
the given data, the resultant profile is presented as the 2nd axis as is outside of the IEC confines, a descriptive warning message is
seen in Fig. 6. printed to the program terminal.

Table 4
Profile parameters.

Parameter name [unit] Explanation

CAX dev [cm] Offset between computed CAX and data position axis 0-point. Computed CAX is derived from the computed field size as
the symmetric center
Sym [%] Profile symmetry is defined inside the flattened region as max(d(x)/d(-x)), where d(x) refers to the dose value at point x.
Dmax [%] Relative maximum dose to CAX within the flattened region
Dmin [%] Relative minimum dose to CAX within the flattened region
Dev [%] Dose deviation between CAX and the minimum dose within the flattened region
Fieldsize [cm] Fieldsize is defined at 50% dose value around the CAX
PenR & PenL [cm] Right and left side penumbras, defines as the distance between 80% and 20% dose values from CAX

Table 5
PDD parameters.

Parameter name [unit] Explanation

R100, R80, R50 [cm] 100%, 80% and 50% relative to maximum dose depths from the surface, respectively.
D100, D200 [%] Relative doses at depths of 10 and 20 cm from the surface, respectively.
Ds [%] Surface dose relative to the maximum dose. Surface is defined as the depth of 0.05mm in the IEC 60976.
J100/J200 Dimensionless quality parameter. Ratio between doses at 10 and 20 cm, respectively

5
T. Pakarinen and J. Ojala Computer Methods and Programs in Biomedicine 212 (2021) 106457

Fig. 6. Quick profile tool for fast profile inspection.

3.4.2. Gamma, DTA and DD analysis Gamma, DTA and DD results are shown automatically in the
Gamma, DTA and DD analysis are performed after the user has main figure legend after computation. The legend is located by de-
set required parameters in the UI panel f). First, the user selects fault in the north-east corner of the figure. For 1D data, all cho-
the data (1D or 2D), so that the reference data is selected first. sen line profiles and pass percentages are presented in the legend,
Gamma criteria are set as default to 3% (DD) and 3 mm (DTA), i.e. target data, reference data, brute force computation results and
but the values can be modified by the user preceding the gamma analytical computation results consisting of DTA, relative DD and
analysis from the respective edit fields. Gamma interpolation res- absolute DD.
olution defaults to the lower resolution data by default, but can 2D pass percentages are presented similarly in the main fig-
also be changed from the ‘interp-interval’ edit field, located in the ure legend, but 2D distributions can only be viewed when refer-
UI group e), which will perform linear interpolation using the re- ence data (data which was chosen 1st to the analysis) is activated.
quested resolution. The gamma, DTA and DD computations are ex- Gamma, DTA or DD distribution can be chosen from UI group b)
ecuted for the whole data area by default, but the user may change from respectively named buttons. Additionally, the quick profile
the analysis area by setting an isodose level from isodose edit field, tool can be used to inspect the results against the dose profile in
which limits the computation to doses equal, or higher to the set 1D.
level as presented in Fig. 7.

Fig. 7. Isodose delineated DD distribution.

6
T. Pakarinen and J. Ojala Computer Methods and Programs in Biomedicine 212 (2021) 106457

Fig. 8. Quick profile comparison between the Eclipse treatment plan and Monte carlo simulated dose data, before gamma analysis.

Fig. 9. Gamma distribution inspection with the quick profile tool.

Gamma, DTA and DD results can be exported to .mat or .csv for- function and constants were first added in a custom function ASCII
mat from the export buttons shown in Fig. 5. .mat format contains file, which was then operated on the 3D dataset via a custom func-
all data in the saved structure including the results, but the .csv tion tool in Profeel (GUI group C). Next, proper transversal slices
format exports only the 1D normalization and interpolated data for analysis were searched, inspected with the quick profile tool
with the results. For 2D data, only final gamma, DTA and DD pass and extracted from both datasets. The quick profile inspection is
percentages are exported during .csv export. presented in Fig. 8
No additional normalization was chosen, since in this case
proper dose calibration data was available. Next 2D gamma anal-
3.5. Sample run
ysis was performed for the data with 20 Gy isodose limit and the
default 3 mm DTA and 3% dose difference criteria. With this con-
Sample run was performed with a DICOM RT export of volu-
figuration, the 2D gamma pass percentage was found to be 94.35%
metric modulated arc (VMAT) plan from Eclipse TPS (version 13.6)
with 79,9% DTA and 60.9% dose difference pass percentages. The
(Varian Medical Systems Inc., Palo Alto, CA, USA) and Monte Carlo
individual gamma and gamma component maps were then in-
simulated .3ddose data for the same case. After import, the units
spected with the quick profile tool, which overlays the gamma re-
of the .3ddose data set were scaled using known calibration in-
sults with the dose distribution as shown in Fig. 9.
formation for the simulated treatment machine. The dose scaling

7
T. Pakarinen and J. Ojala Computer Methods and Programs in Biomedicine 212 (2021) 106457

Fig. 10. Gamma analysis results for 1D comparison.

use is to import calculated dose distribution from a clinical TPS

and independent Monte Carlo simulation result and measured dose
distribution. The data can be visualized and also lower dimensional
(3D to 2D or 2D to 1D) sub-data from the original data can be ex-
tracted. This is essential, since normally the dose distributions are
calculated in 3D, but the measurements and thus the data analy-
sis and comparison are performed in 2D or 1D. There are multi-
ple options for data processing, including smoothing, analysis and
comparison tools. The simplest comparison is to evaluate the dose
difference, but the more complex the dose distribution, more ad-
vanced metrics, such as gamma analysis are favoured. Finally, the
data can be saved or exported for further use. Support for new
data formats, tools for data processing and analysis/comparison
and other options in the software can be added through improve-
ment requests in Profeel GitHub repository.
In future, Profeel continues supporting medical physicists and
scientists in dosimetry analysis and research. Thus far Profeel has
shown to be an agile tool for fulfilling various needs of several re-
searchers and since Profeel is under constant development and as
an open source project, community needs, issues and bug reports
are taken into account in the development. Profeel is not intended,
and should not be used in clinical practice.
Fig. 11. Exported 1D analysis results in .csv form.
There are several areas that should be covered in future devel-
opment, such as a broader range of importable data types/formats,
After the inspection, limited range Y-axis 1D profiles were fur- e.g. other Monte Carlo codes, more advanced analysis tools, espe-
ther extracted and both data were filtered with Savitzky-Golay fil- cially in 3D (e.g. DVH calculation, radiobiological modelling), CT
ter (UI group C), with 3th polynomial order for 6 frame window and MRI image import with dosimetric data registration, exten-
size. Filtered profiles with gamma results are shown in Fig. 10. sion of gamma analysis and Savitzky-Golay smoothing filter to 3D,
Finally, the results were exported as .mat and in .csv form. DICOM-RT structure import and overlay and extensive program
Fig. 11. presents a typical .csv export file. verification and validation. Also, spline interpolation for gamma
analysis and gamma parameter reports should be included as de-
4. Discussion scribed by Hussein et al. [16]. Additional known limitation is the
Matlab GUIDE environment, which will become obsolete in future
Profeel is a software, where 1D/2D/3D dose distributions, Matlab releases. Profeel can be still run in newer Matlab versions
whether calculated or measured, of multiple formats can be im- but to edit UI tools, the GUI must first be migrated to Matlab App
ported, processed and analyzed. Number of free-to-use and open Designer (MathWorks Inc., MA). The specific instructions to run
source programs are focusing on the dosimetric analysis in radio- Profeel, with required Toolboxes are described in the Profeel docu-
therapy, but unfortunately the comparison between measured and mentation [14]. All known program bugs and issues are reported to
simulated data is often inaccessible without extensive preprocess- the Profeel Github repository at: https://github.com/TPakar/Profeel/
ing. Thus, Profeel focuses mainly on this issue. Typical example of issues.

8
T. Pakarinen and J. Ojala Computer Methods and Programs in Biomedicine 212 (2021) 106457

Acnowledgemenets [7] Sang-Won Kang, Jin-Beom Chung, Kyeong-Hyeon Kim, Ji-Yeon Park,
Hae-Jin Park, Woong Cho, Sven Olberg, Tae Suk Suh, Justin Park, Devel-
opment of Volumetric Independent Dose Calculation System for Verification
Authors have no competing interest to declare. This research of the Treatment Plan in Image-Guided Adaptive Brachytherapy, Frontiers in
did not receive any specific grant from funding agencies in the Oncology (2020) 10.609.10.3389/fonc.2020.00609.
public, commercial, or not-for-profit sectors. [8] Martinov, M., Thomson, M. 3ddose_tools, v. 1.1, GitHub repository. 2016.
Retrieved from https://physics.carleton.ca/clrp/3ddose_tools/3ddose-tools [Ac-
cessed 28.4.2021].
References [9] Badum, E. VICTORIA - Voxel Interactive Contour Tool for Online Radiation
Intensity Analytics. 2021. Retrieved from http://web.uvic.ca/∼bazalova/dose_
[1] J. Deasy, A. Blanco, V. Clark, CERR: a computational environment for radio- viewer [Accessed 28.4.2021].
therapy research, Med Phys 30 (5) (2003 May) 979–985 PMID: 12773007, [10] DA Low, WB Harms, S Mutic, JA. Purdy, A technique for the quantitative eval-
doi:10.1118/1.1568978. uation of dose distributions, Med Phys 25 (5) (1998) 656–661, doi:10.1118/1.
[2] C Pinter, A Lasso, A Wang, D Jaffray, G. Fichtinger, SlicerRT: radiation therapy 598248.
research toolkit for 3D Slicer, Med Phys 39 (10) (2012 Oct) 6332–6338 PMID: [11] Geurts, M. CalcGamma, GitHub repository. 2014. Retrieved from https://github.
23039669, doi:10.1118/1.4754659. com/mwgeurts/gamma [Accessed 5.8.2020].
[3] Omer, H., Alzimami, K. and Sulieman, A., “XSTING, Software for Evaluation [12] YJ Graves, X Jia, SB. Jiang, Effect of statistical fluctuation in Monte Carlo based
of Radiotherapy Planning Performed Using EGSnrcMP Codes”, Current Medical photon beam dose calculation on gamma index evaluation, Phys Med Biol 58
Imaging. 2014 10: 48. doi: 10.2174/157340561001140424145206. (6) (2013 Mar 21) 1839–1853 Epub 2013 Feb 27. PMID: 23442443, doi:10.1088/
[4] JY Park, JW Lee, KS Choi, JS Lee, YH Kim, S Hong, TS Suh, Development of a 0031-9155/58/6/1839.
novel quality assurance system based on rolled-up and rolled-out radiochromic [13] MathWorks. 2020. Matlab Documentation - Savitzky-Golay filtering. Re-
films in volumetric modulated arc therapy, Med Phys 38 (12) (2011 Dec) 6688– trieved from https://se.mathworks.com/help/signal/ref/sgolayfilt.html [Ac-
6696 PMID: 22149851, doi:10.1118/1.3659706. cessed 9.4.2021].
[5] H Wang, Q Liu, D Wan, J Xiang, L Du, Y Wang, J Cao, Y Fu, F Fan, M. Hecker, [14] T. Pakarinen, J.Profeel Ojala, Open Source Dosimetry Data Visualization and An-
BioDoser: improved dose-estimation software for biological radiation dosime- alytics Software, v. 1.17, GitHub repository (2020). [Accessed 9.4.2021]. https:
try, Comput Methods Programs Biomed 108 (1) (2012 Oct) 402–406 Epub 2012 //github.com/TPakar/Profeel.
Apr 14. PMID: 22503129, doi:10.1016/j.cmpb.2012.03.010. [15] IEC - International Electrotechnical Commission, Medical electrical equipment
[6] H.-P. Wieser, E. Cisternas, N. Wahl, S. Ulrich, A. Stadler, H. Mescher, L.-R. Müller, - Dosimeters with ionization chamber as used in radiotherapy, IEC Standard
T. Klinge, H. Gabrys, L. Burigo, A. Mairani, S. Ecker, B. Ackermann, M. Eller- 60976 (2011) Genève: IEC.
brock, K. Parodi, O. Jäkel, M Bangert, Development of the open-source dose [16] M Hussein, CH Clark, A. Nisbet, Challenges in calculation of the gamma index
calculation and optimization toolkit matRad, Med. Phys. 44 (2017) 2556–2568, in radiotherapy - Towards good practice, Phys Med 36 (2017 Apr) 1–11 Epub
doi:10.1002/mp.12251. 2017 Mar 14. PMID: 28410677, doi:10.1016/j.ejmp.2017.03.001.