Escolar Documentos
Profissional Documentos
Cultura Documentos
Variations in the Contouring of Target Structures and Organs at Risk: Test Case from a
Patient with Intermediate Risk Prostate Cancer
Bijoy Anand, M.S., Brittany Bird, B.S., R.T.(R)(T), Seth Cox, B.S.,
Ashley Hunzeker, M.S., CMD, Veronica Laird B.S., R.T.(T),
Nishele Lenards, M.S., CMD, R.T.(R)(T), FAAMD, Stephanie Sanford, MSHA, R.T.(R)(T),
Pat Sheil, B.S., R.T.(T), Ashley Smelko B.S., R.T.(T), Sadie Wilhite, B.S., R.T.(R)
Abstract:
Introduction
A growth in technology has changed the spectrum of radiation oncology and has created
the ability to irradiate and target specific tumor volumes while sparing organs at risk (OR) via
contours drawn by medical professionals. Volumetric-modulated arc therapy (VMAT) and
intensity-modulated radiation therapy (IMRT) are two specific techniques that allow for
increased precision, conformity, and achieve a steep dose gradient to spare nearby critical
structures. The popularity of these techniques compared to 3D conformal radiotherapy has
grown immensely; yet if the structures are not clearly defined, the overall plan quality will be
affected. In addition to treatment technique, anatomical contouring is an essential part of the
treatment planning process. Contours create dose tracking not only to target volumes but also to
OR; however, incorrect contours or variances within contours can create inaccurate dose
administration and tracking. Deviations within contours may lead to over irradiation of normal
tissue or cause under-dosing to portions of the target volume.1
A select variety of contouring applications are available that allow the user to accurately
project the volume of an organ or delineate a target structure, interpolate between image slices,
define margins, and much more. The tools inlaid in the contouring application allow the user to
perform the necessary tasks prior to beginning the treatment planning process. While contouring
is an important component, variations can inadvertently reflect a false depiction onto the dose-
volume histogram (DVH). There has been a lack of quality applications that can measure or
quantify the consistency and accuracy of contours, until most recently with the introduction of
ProKnow. ProKnow is one of the first analytic companies to develop a software program for
measuring contouring and radiation accuracy. In-depth studies identifying metrics of contouring
discrepancies can facilitate more applications to be launched and implemented into the field of
dosimetry, resulting in a reduction of dosimetric impact.
2
standard contours are shown in Figure 1. The guidelines specified in the Radiation Therapy
Oncology Group (RTOG) 0126 were utilized for contouring, in particular, the following
definitions:4
Rectum: Inferiorly from the lowest level of the ischial tuberosities (right or left).
Contouring ends superiorly before the rectum loses its round shape in the axial plane and
connects anteriorly with the sigmoid.
Bladder: Inferiorly from its base, and superiorly to the dome.
Prostate: Inferiorly from its apex and superiorly to its base. If the capsule is visible, the
muscles and soft tissues abutting the capsule are not included as 'prostate'.
Seminal vesicles: Entire seminal vesicles including those slices that also have prostate
identified.
Penile bulb: The portion of the bulbous spongiosum of the penis immediately inferior to
the GU diaphragm. Structure not extended anteriorly into the shaft or pendulous portion
of the penis.
Treatment Planning
A treatment plan was generated using Pinnacle3 version 14.0 treatment planning system
(TPS) on the gold standard contour set. The prescription dose was set to 180 cGy per fraction for
39 fractions for a cumulative dose of 7020 cGy. An Elekta Agility linear accelerator equipped
with a 160 multi-leaf collimator was selected as the beam delivery device. The dose
computation method was the adaptive convolution algorithm. A total of two 10 MV VMAT
beams were chosen for this treatment plan given the depth of the target volume. Each beam
consisted of full arcs with the clockwise (CW) beam rotating from 179 to 181, and the counter-
clockwise (CCW) beam rotating from 180 to 178. A collimator angle of 5 was set on the CW
arc and a collimator angle of 355 was set on the CCW arc to avoid overlapping interleaf
transmission. The CW beam contributed 48.12% of the dose while the CCW contributed
51.88%. The dose grid was set to a 0.3 x 0.3 x 0.3 cm resolution for optimization.
The plan was produced by inverse planning and by applying constraints from arm 1 of
the RTOG 0126 study.1 The calculated monitor units (MUs) for this plan had a total of 290.3
MU per fraction. The global maximum dose was 7387 cGy, or 105% of the prescription dose.
Figure 2 demonstrates the isodose distribution for this treatment plan.
Plan Comparisons
4
Several metrics were utilized for a comparison of each set of contours against the plan
created and optimized utilizing the gold standard contours. The items that were evaluated are
listed in Table 1 of this article. The dosimetric effects due to contouring differences were
evaluated by replacing the gold standard contours with each clinician's contours, while keeping a
fixed 3D dose grid.
Results
greater than the gold standard mean dose and the bladder mean dose varying by 3.9% greater
than the gold standard mean dose.
The 13 different metrics analyzed among the 23 datasets obtained can be found in Table
3. Of particular interest were the PTV coverage and maximum dose to the PTV as they
accounted for the highest weighting in scoring (16 pts/each). Volume (%) of the PTV covered
by 70.2 Gy had two goals: 95% being the minimum requirement and 100% being the ideal
requirement. Out of the 23 datasets, 15 failed to meet the minimum requirement and all 23 failed
to meet the ideal requirement. The mean result among all datasets was 94.30%. Volume (%) of
the PTV covered by 75.1 Gy had two goals: 2% being the minimum requirement and 0% being
the ideal. All 23 datasets passed the minimum and ideal goal of this metric. Maximum dose (Gy)
to the PTV had two goals: 77.2 Gy being the minimum requirement and 75.1 Gy being the ideal.
All 23 datasets passed the minimum and ideal goal for this metric as well.
The maximum possible metric score for all 13 sections combined was 116 pts. The total
metric scores for all 23 datasets were added and divided by the cumulative score to find the mean
score. The mean score amongst the entire data collected was 97.9/116 (84.4%).
Finally, a comparison of organ volume variation over all clinicians was done. Taken into
consideration the 23 datasets utilized, profession did not show any difference in volumes of
contours created. 14 medical dosimetrists, 4 physicians and 3 physicist contoured structure sets.
The variance between each profession is presented in Table 4. Dosimetrist had the lowest
variance in mean dose of rectum within their planning, with an average mean dose of 31.6 Gy, or
a difference of 2.5%, while physicist had the largest variance in mean dose of the rectum at 34.4
Gy or 11.6%. On the other hand, physicians had the best mean dose for the bladder at 49.2 Gy or
1.8% while the dosimetrist had the greatest variance at 50.5 Gy or 4.5%.
Maximum dose is a completely different result though, as the maximum dose for each
profession is within 0.5% of the gold standard maximum dose for the rectum and within 0.4% of
the gold standard maximum dose for the bladder.
All of the professions struggled with the penile bulb mean dose. Dosimetrist were the
closest to fall in line with the mean dose at 23.33 Gy, or 76% higher than the gold standard.
However, physicists' structure sets received 42.52 Gy mean dose delivered to the penile bulb, a
6
221% greater dose than the gold standard. While the mean dose is still within the DVH criteria,
it is a significant difference. PTV mean dose, however, was close for all professions, all within
0.5% of the gold standard set.
Discussion
This study observed the variation of contouring of OR and target volumes over a
population of medical dosimetrists, physicists, and radiation oncologists for an intermediate risk
prostate cancer CT dataset. It is obvious that there are many variations of inter-clinician
contouring of the male pelvis. The bladder proved to have the least variability. The increased
contouring accuracy of the bladder is expected due to its distinct boundaries that are often easily
identified on a CT scan.1 The rectum had the next lowest variability, following the bladder. The
superior portion of the rectum often has well defined borders. Moving inferiorly, the rectum is
commonly more difficult to identify as it reaches the prostate boundary and the anal canal. 5 The
penile bulb may only be seen on a few slices of the CT scan, making it difficult to distinguish.
The variation of volume can play a huge role in the evaluation of the plan. Variation in
OR volume can drastically change the maximum dose to that structure and will also play a role
in the increase or decrease of mean dose. When it comes to contouring OR, contouring accuracy
is extremely important in dose tracking. If voxels are missing, the dose to that area is not being
recorded. This may be an area of maximum dose that is being missed for that structure. For the
OR, over-contouring is not as crucial, but it may produce an inaccurate mean dose. If the extra
voxels are contoured outside of the area of high-dose, then the mean dose will be lower than it
should be. It would have the opposite affect if the extra voxels were contoured within the area of
high-dose.3 If the target structure is over-contoured, then the dose may increase to the
surrounding OR in order to get dose coverage of those extra voxels. If the target structure is
under-contoured, then there may be disease outside of the treated area that is not being
adequately covered.
In order to compare the inter-clinician contouring variability, there must be a set of
reference contours created for the dataset on which a treatment plan is created to meet the
required objectives. The gold standard reference contours may be of question when considering
the results of this study. For this study, the gold standard contours are used as a standard to
compare the other clinicians contours to. We must keep in mind that the goal of this study is to
7
observe any contour variations and the impact of the variations. Any set of contours could have
been used as the gold standard reference set of contours, but there still would have been
variability between each of the clinicians. The important fact is to notice that contouring
variability does exist and it does affect the DVH results.
One of the aspects that often is brought up is experience and training. There is a small
correlation between this in terms of mean and maximum doses. However, the greatest variance
was for medical physicists given that their mean rectal dose resulted significantly higher. This
could be attributed to the lack of rectal contouring physicists perform on a daily basis, but also
could be due to very small sample size of population within the study - only three respondents
were physicists. The overall results show there is no correlation between OR mean and
maximum dose and level of education.
Another significant point was that penile bulb contouring had the largest variation.
Segmentation of the penile bulb should be discussed and standardized throughout this profession,
as each medical dosimetrist, physicist and radiation oncologist all had differing volumes, and
consequently, varying mean doses throughout.
Conclusion
References
1. Collins KS. An evaluation of the contouring abilities of medical dosimetry students for
the anatomy of a prostate cancer patient. Med Dosim. 2012;37(3):245-249.
http://dx.doi.org/10.1016/j.meddos.2011.09.003
2. Bortfeld T, Jeraj R. The physical basis and future of radiation therapy. Br J Radiol.
2011;84(1002):485-498. http://dx.doi:10.1259/bjr/86221320
3. Nelms BE, Tom WA, Robinson G, Wheeler J. Variations in the contouring of organs at risk: test case
from a patient with oropharyngeal cancer. Int J Radiat Oncol Biol Phys. 2012;82(1):368-378.
http://dx.doi.org/10.1016/j.ijrobp.2010.10.019
4. Michalski, Jeff. "A Phase III Randomized Study of High Dose 3DCRT/IMRT versus
Standard Dose 3DCRT/IMRT in Patients Treated for Localized Prostate Cancer." RTOG,
n.d. Web. 27 July 2017.
https://www.rtog.org/ClinicalTrials/ProtocolTable/StudyDetails.aspx?study=0126.
5. Barghi A, Johnson C, Warner A, Bauman G, Battista J, Rodrigues J. Impact of
contouring variability on dose-volume metrics used in treatment plan optimization of
prostate IMRT. Cureus. 2013;5(11). http://dx.doi.org/10.7759/cureus.144
9
Figures
Figure 1. Gold standard contours of the bladder (yellow), penile bulb (pink), rectum (green),
CTV (blue), and PTV (red).
10
Figure 2. The isodose distribution on the gold contour treatment plan. The bladder is contoured
in yellow and rectum is contoured in green. The light green isodose line is the 100% line, which
encompasses the PTV (color washed red).
11
Table 2. Objectives for all of the planning volumes and OR were met or exceeded when
comparing constraints to the RTOG 0126 study.
Structure Dose Volume Achieved Goals
Constraint Constraint Met?
PTV 110% (7720 2% (minor 0%
cGy) variation)
PTV 107% (7510 2% 0%
cGy)
PTV 100% (7020 98% 98.05%
cGy)
CTV 100% (7020 100% 100%
cGy)
Bladder 8000 cGy <15% 0%
Bladder 7500 cGy <25% 0%
Bladder 7000 cGy <35% 18.75%
Bladder 6500 cGy <50% 29.63%
Rectum 7500 cGy <15% 0%
Rectum 7000 cGy <25% 0.61%
Rectum 6500 cGy <35% 2.59%
Rectum 6000 cGy <50% 4.19%
Penile Bulb Mean Dose 1314 cGy
5250 cGy
13
Figure 3. Histograms of OR volumes defined for the same CT image set (defined by 23
contouring clinicians per OR). The distribution of OR volume is a simple metric of variability.
14