Realistic Head and Neck Geometry

1
Investigation of Planned Surface Dose Accuracy
and Obliquity Factorin Head and Neck Intensity
Modulated Radiation Therapy (IMRT)K Hirmiz, L
Mulroy, D Wilke, M Rajaraman, T Connell and J
Robar Nova Scotia Cancer Centre and Department
of Radiation Oncology, Dalhousie University,
Halifax, Nova Scotia, Canada
Table 2
Table 1
Slab geometry dose accuracy in the build-up
region
The anthropomorphic phantom study described above
provides a realistic treatment geometry with
multiple IMRT ports, but does not allow for
variation of measurement depth below the surface.
For this we used the phantom geometry shown in
Figure 5, consisting of a Capintec PS-033
parallel plate chamber embedded within a
polystyrene slab, below a selectable depth of
polystyrene buildup. Measurements were made using
each of the 7 ports for the first head and neck
IMRT patient treated in Atlantic Canada. For
each port, measurements were made on central axis
for depths of 1.2, 1.5, 2.1, 2.0, 3.1, 3.5, 4.3,
5.0, 10.0, 15.0, 20.0, and 50.0 mm. The gantry
was set to 0? for all measurements.   Using the
planning system, the two-dimensional absolute
dose distribution in a tissue-equivalent slab was
exported for each IMRT port at the depths
corresponding to the measurements. External
software (Figure 6) was developed (MatLAB,
Mathworks, Natick, MA) to read each distribution
and to calculate the mean dose value in the area
occupied by the parallel plate chamber. Planned
dose values thus determined were compared with
measured values.
Treatment of the head and neck is a widespread
application of Intensity Modulated Radiation
Therapy (IMRT)1. Compared to conventional
methods, IMRT offers an improved capacity to
provide adequate dose coverage of complex target
volumes, while sparing multiple organs at risk,
including spinal cord, brainstem, optic nerves
and parotid glands. Commissioning and
patient-specific quality assurance protocols
usually include confirmation of the dose
distribution at depth in phantom using detectors
such as radiographic film and ionization
chambers. However, assessment of the accuracy of
the planned dose in on the skin surface and in
the buildup region is also crucial but less
commonly-performed.     The aims of this present
study are twofold. The first goal is to validate
the surface and buildup doses for head and neck
IMRT treatment, as predicted by our own planning
system (BrainSCAN 5.2, BrainLAB AG). To our
knowledge, this planning system has not been
investigated in terms of buildup dose accuracy
for IMRT treatment planning. This
investigation includes the comparison of planned
and measured surface doses for both phase I and
phase II of a realistic head and neck treatment.
For normal beam incidence, we quantify the
disparity between planned and measured doses as a
function of buildup depth. The second goal of
this work is to assess the importance of IMRT
port obliquity in determining the surface dose as
compared to non-modulated ports by studying the
variation of obliquity factor with angle of port
incidence.
Discussion
The practical goal of this work was to determine
whether planned and measured surface and
buildup-region doses were in agreement for IMRT
treatment of the head and neck in our clinic.
Over the course of an extensive commissioning
process, it had been determined that that
measured and planned absolute dose agree at depth
(e.g. isocentre) within ?2. However, from the
micro-MOSFET measurements on the anthropomorphic
phantom, it was observed that the planning system
overestimates the dose at the surface
significantly (on average, by a factor of 1.4 and
1.6 for phase I and II, respectively). Examining
this dose agreement as a function of depth
clarifies the situation this disparity does not
persist over a large depth, and in fact the
planned and measured dose begin to agree
reasonably at approximately 2 mm below the
surface for most IMRT ports. For most ports
there is a underestimation of the dose by the
planning system at depths between 2 mm and 3.5
mm. However, with reference to Figure 7, this
underestimation is simply the result of a small
spatial mismatch in the depth dose curves in this
high-dose-gradient region. In this region, the
distance-to-agreement is less than 1 mm and thus
within our usual spatial tolerance (3 mm). A
similar observation of overestimated planned
surface dose in IMRT has been made by Dogan et
al5, while Mutic and Low7 have observed the same
pattern of overestimation at the surface,
followed by underestimation of dose several
millimeters below the surface. More recently
Chung(10) et al observed an overestimation of the
surface and build-up dose by the planning system
by 7.4 and 18.5 respectively. IMRT ports
exhibit a different dependence of surface dose on
port obliquity compared to open ports with the
same outer aperture. Whether the dependence is
greater than or less than that of an open port
apparently depends on the direction of the local
fluence gradient in the region of central axis
relative to the direction of gantry rotation.
This is in keeping with the Electron Range
Surface (ERS) explanation (4) of increase of
surface dose with beam obliquity. For IMRT
ports, a higher fluence towards the acute angle
of beam incidence (i.e. the port region which
impinges on the phantom to a greater extent with
obliquity) would increase (to a greater extent
relative to a lower fluence region) the
contribution of electrons to the central axis
surface location. This observation warrants
further investigation with simplified IMRT
sequences and Monte Carlo modeling.
Obliquity factor IMRT versus open ports
To study the variation of surface dose with port
obliquity, the arrangement shown in Figure 5 was
used without buildup (i.e. d0 mm). For each of
the seven IMRT ports used in phase 2 of the
anthropomorphic phantom plan, separate
measurements were made with gantry angles ranging
from 0? to 70? from the normal. Obliquity
Factor, OF(?), was calculated as the ratio of the
surface dose at a given gantry angle ? to the
surface dose for normal incidence. For
comparison, this measurement was repeated using
the non-modulated Complete Irradiated Area
Outline (CIAO) for each port. All readings were
corrected for the variation of detector response
with beam obliquity according to the data
provided for the PS-033 chamber by Gerbi and
Khan.6
Realistic Head and Neck Geometry
The aim of this first study was to simulate the
patient imaging, IMRT treatment planning and
delivery processes using an anthropomorphic
phantom, and to compare planned and delivered
surface dose at the entry point of each IMRT on
the surface of the phantom. The head and neck
section of the Rando anthropomorphic phantom
(Phantom Laboratory, NY) was imaged using CT.
Simulated volumes including the GTV, CTV and PTV
were contoured by the radiation oncologist for
phase 1 (gross tumour and elective nodal volume)
and phase 2 (gross tumour boost). Organs-at-risk
were contoured, including the spinal cord,
brainstem, optic chiasm, optic nerves, lenses and
both parotids (Figure 1). For each treatment
phase, separate IMRT plans were generated. For
both phases, seven coplanar ports were used as
shown in Figure 2 to mimic our standard class
solution for head and neck. MLC sequencing was
performed for segmental treatment delivery, and
on average, each port consisted of 30 segments.
After the completion of treatment planning, the
surface doses at the central axis entry point of
each port were recorded (Figure 3) with a spatial
precision of approximately ?1.0 mm. For both
phases, the planned dose per fraction at
isocentre was 2.0 Gy.   For both treatment plans,
the phantom was aligned on the linear accelerator
couch, and micro-MOSFETS (Thomsen Neilson,
Ottawa) were attached at the central axis entry
points of each port (Figure 4) using the light
field cross-hairs. The micro-MOSFETS have bulb
dimensions of 1.0 mm ? 3.5 mm. Following
treatment delivery of all ports (to give the
planned 2.0 Gy at isocentre) the readings from
all micro-MOSFETs were obtained.
Realistic Head and Neck Geometry Table 1 gives
the comparison of measured and planned surface
doses for the phase 1 treatment of the
anthropomorphic phantom. For this treatment
plan, the expected surface dose was expected to
be high (54 to 95 of the dose at isocentre
depending on the measurement location), given the
extension of the PTV to the surface. The
planning system overestimates the surface dose by
a factor of 1.2 to 1.59 (mean 1.40). Table 2
gives a similar comparison for the phase 2
treatment (where surface doses were expected to
be comparatively low to achieve skin spaing),
indicating that the planned dose was
overestimated by a factor of 1.24 to 2.41 (mean
1.66).     Systematic study of agreement of
planned and measured dose in build-up
region Figure 7 compares the planned versus
measured absolute dose as a function of depth in
phantom, for the first of the 7 ports studied.
Consistent with the anthropomorphic phantom
study, the planning system overestimates the
surface dose significantly.   Figure 8 gives
the ratio of planned to measured relative dose as
a function of depth the same port. A
normalization depth of 50 mm was used following
verification that planned absolute dose values at
this depth were within 2 of measured values.
For all ports a consistent trend was observed
over the first 2.0 mm depth on average, the
planning system overestimates the relative dose
significantly, by as much as 30. Significant
discrepancy is resolved at a depth of 2.0 mm.
Between 2.0 and 3.5 mm depth, the planning system
underestimates the delivered relative dose by
approximately 4 on average. Beyond a depth of
3.5 mm, for five of the seven ports studied, the
agreement is within 2. Beyond 15 mm (dmax), the
agreement is within our expected tolerance (?3).
Validation of delivered surface dose in head and
neck IMRT is essential to ensure either coverage
of superficial target volumes or skin sparing.
Due to the complexity of the mechanisms
underlying dose deposition in the build-up
region, it is understandable that the accuracy of
most convolution-based forward dose calculations
will be limited. The IMRT planning system used
in this work overestimates the surface dose, by a
factor of 1.4 to 1.66 on average, for phase I and
phase II plans studied. However, this
discrepancy is resolved over the first 2.0 mm
depth in tissue. The implication of this finding
is that the dose will be sufficiently accurate at
the location of superficial nodal target volumes.
In situations where skin is felt to be involved
by tumour and is in the the high dose target
volume, it would be prudent to take addition
measures, e.g., addition of bolus, to ensure
adequate dose coverage in this region. Measured
results show that the obliquity factor for IMRT
ports may differ significantly from open ports
with the same outer aperture (by a ratio ranging
from 0.79 to 1.17 for the ports studied.)
Whether the OF is greater or less than that for
an open port apparently depends on the direction
of fluence gradient across the port relative to
the direction of gantry rotation.

• 1. Mell LK, Roeske JC and Mundt AJ A survey of
  intensity-modulated radiation therapy use in the
  United States. Cancer 200398(1)204-211.
• 2. Yokoyama S, Roberson PL, Litzenberg DW, Moran
  JM and Fraass BA Surface buildup dose dependence
  on photon field delivery technique for IMRT. J
  Appl Clin Med Phys 20045(2)71-81.
• 3. Lee N, Chuang C, Quivey JM, et al. Skin
  toxicity due to intensity-modulated radiotherapy
  for head-and-neck carcinoma. Int J Radiat Oncol
  Biol Phys 200253(3)630-637.
• 4. Khan FM. The Physics of Radiation Therapy.
  Baltimore Lippincott, Williams and Wilkens
  1994.
• 5. Dogan N and Glasgow GP Surface and build-up
  region dosimetry for obliquely incident intensity
  modulated radiotherapy 6 MV x rays. Med Phys
  200330(12)3091-3096.
• 6. Gerbi BJ and Khan FM Plane-parallel
  ionization chamber response in the buildup region
  of obliquely incident photon beams. Med Phys
  199724(6)873-878.
• 7. Mutic S and Low DA Superficial doses from
  serial tomotherapy delivery. Med Phys
  200027(1)163-165.
• 8. Gerbi BJ, Meigooni AS and Khan FM Dose
  buildup for obliquely incident photon beams. Med
  Phys 198714(3)393-399.
• Jackson W Surface effects of high-energy X rays
  at oblique incidence. Br J Radiol
  197144(518)109-115.
• Chung H Evaluation of surface and build-up region
  dose for IMRT in head and neck cancer.Med
  Phys.32(8), 2682-2689.

Variation of surface dose with beam obliquity Our
measurements show that the obliquity factor (OF)
for IMRT ports may differ significantly from an
open (i.e. non-modulated) port with the same
outer outer aperture. Figures 9 and 10
illustrate this comparision for two of the seven
ports for the phase 2 treatment. It is noteable
that the OF for the IMRT port may be greater or
less than that of the non-modulated port. Among
all seven ports studied, we observed a ratio
OFIMRT/OFopen ranging from 0.79 to 1.17. The
fact that the dependence of surface dose on
obliquity for a IMRT port may be more or less
pronounced than that for an open port is
apparently related to the fluence gradient across
the port relative to the direction of gantry
rotation. If the fluence is greater (relative to
central axis) on the side of the port towards the
acute angle of incidence (i.e., the region of the
port that will be incident on the phantom at
shorter distance from the source), the OF for the
IMRT port will show a greater increase with
gantry angle compared to an open port.
ASTRO 2005, Denver CO