International Journal of

Radiation Oncology
biology physics

www.redjournal.org

Physics Contribution

Assessment of Interfraction Patient Setup for

Head-and-Neck Cancer Intensity Modulated Radiation
Therapy Using Multiple Computed Tomography-Based
Image Guidance
X. Sharon Qi, PhD,* Angie Y. Hu, PhD,y Steve P. Lee, MD, PhD,* Percy Lee, MD,*
John DeMarco, PhD,* X. Allen Li, PhD,z Michael L. Steinberg, MD,*
Patrick Kupelian, MD,* and Daniel Low, PhD*
*Department of Radiation Oncology, David of Geffen School of Medicine at UCLA, Los Angeles, California; yDepartment of
Radiation Oncology, University of Colorado School of Medicine, Aurora, Colorado; and zDepartment of Radiation
Oncology, Medical College of Wisconsin, Milwaukee, Wisconsin

Received Oct 9, 2012, and in revised form Jan 9, 2013. Accepted for publication Jan 15, 2013

Summary
Image guidance systems are
commonly used with
intensity modulated radiation
therapy (IMRT) in head-and-
neck cancer irradiation. We
retrospectively analyzed
3302 interfraction computed
tomography (CT) images for
117 patients using multiple
image guidance modalities:
kilovoltage cone-beam CT
(KVCBCT), megavoltage
fan-beam CT (MVFBCT),
and megavoltage cone-beam
CT (MVCBCT). Our data
suggest that the clinical
target volume-to-planned
target volume (CTV-to-PTV)
margin for head-and-neck
IMRT may be a function of
Purpose: Various image guidance systems are commonly used in conjunction with intensity
modulated radiation therapy (IMRT) in head-and-neck cancer irradiation. The purpose of this
study was to assess interfraction patient setup variations for 3 computed tomography (CT)-based
on-board image guided radiation therapy (IGRT) modalities.
Methods and Materials: A total of 3302 CT scans for 117 patients, including 53 patients
receiving megavoltage cone-beam CT (MVCBCT), 29 receiving kilovoltage cone-beam CT
(KVCBCT), and 35 receiving megavoltage fan-beam CT (MVFBCT), were retrospectively
analyzed. The daily variations in the mediolateral (ML), craniocaudal (CC), and anteroposterior
(AP) dimensions were measured. The clinical target volume-to-planned target volume (CTV-to-
PTV) margins were calculated using 2.5S þ 0.7 s, where S and s were systematic and random
positioning errors, respectively. Various patient characteristics for the MVCBCT group,
including weight, weight loss, tumor location, and initial body mass index, were analyzed to
determine their possible correlation with daily patient setup.
Results: The average interfraction displacements ( standard deviation) in the ML, CC, and AP
directions were 0.5  1.5, 0.3  2.0, and 0.3  1.7 mm (KVCBCT); 0.2  1.9, 0.2  2.4,
and 0.0  1.7 mm (MVFBCT); and 0.0  1.8, 0.5  1.7, and 0.8  3.0 mm (MVCBCT). The
day-to-day random errors for KVCBCT, MVFBCT, and MVCBCT were 1.4-1.6, 1.7, and
2.0-2.1 mm. The interobserver variations were 0.8, 1.1, and 0.7 mm (MVCBCT); 0.5, 0.4, and
0.8 mm (MVFBCT); and 0.5, 0.4, and 0.6 mm (KVCBCT) in the ML, CC, and AP directions,
respectively. The maximal calculated uniform CTV-to-PTV margins were 5.6, 6.9, and 8.9 mm
for KVCBCT, MVFBCT, and MVCBCT, respectively. For the evaluated patient characteristics,
the calculated margins for different patient parameters appeared to differ; analysis of variance
(ANOVA) and/or t test analysis found no statistically significant setup difference in any direction.

Reprint requests to: X. Sharon Qi, PhD, Department of Radiation Conflict of interest: none.
Oncology, University of California, 200 UCLA Medical Plaza, Suite B265,
Los Angeles, CA. Tel: (310) 983-3463; E-mail: xqi@mednet.ucla.edu

Int J Radiation Oncol Biol Phys, Vol. 86, No. 3, pp. 432e439, 2013
0360-3016/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.ijrobp.2013.01.022
Volume 86  Number 3  2013 Interfraction patient setup for head-and-neck IMRT 433

the imaging modality if Conclusions: Daily random setup errors and CTV-to-PTV margins for treatment of head-and-
determined based on image neck cancer were affected by imaging quality. Our data indicated that larger margins were asso-
guided RT (IGRT) data when ciated with MVFBCT and MVCBCT, compared with smaller margins for KVCBCT. IGRT
IGRT is absent. IGRT modalities with better image quality are encouraged in clinical practice. Ó 2013 Elsevier Inc.
modalities with better image
quality are encouraged in
clinical practice.

Introduction a Siemens accelerator. We further correlated the setup errors and

the calculated CTV-to-PTV margins with patient characteristics
such as age, sex, and body weight for the MVCBCT patients.
Accurate treatment dose delivery is of particular importance for
patients with head-and-neck (H&N) cancers because of the
proximity of tumor volumes to critical structures (ie, spinal cord) Patient characteristics in MVCBCT group
and the sharp dose falloff of current delivery techniques. Since the
late 1990s, intensity modulated radiation therapy (IMRT), which
delivers highly conformal dose distributions to the tumor while Our largest group in this study, the MVCBCT group, including 30
sparing adjacent normal structures, has become the standard men and 23 women, were further analyzed for correlation between
treatment for head-and-neck cancers (1, 2). To meet specific dose setup errors and patient characteristics. The ages were between 22
constraints for both planned tumor volume (PTV) coverage and and 93 years, and the initial weights ranged from 87.4 to 278.6
critical structure sparing, IMRT generates rapid dose falloff. PTV pounds. During the treatment course, patient weights were
margins and reproducible setups are used to ensure that the target measured weekly, the measured changes (loss) ranged from 1.6% to
gets the prescribed dose. 22.7% (mean weight loss, 7.9%). The MVCBCT patient
Various image guided radiation therapy (IGRT) techniques
have been developed and used to ensure accurate interfractional
patient setup and repeatable dose delivery. The commercially Table 1 Patient demographics in MVCBCT group
available in-room computed tomography (CT)-based imaging Variable No. of patients
systems include kilovoltage cone beam CT (KVCBCT), mega- No. of patients 53
voltage cone-beam CT (MVCBCT), and kilovoltage and mega- Sex
voltage fan-beam CT (KVFBCT and MVFBCT) (3-5). By M 30
collecting daily CT images immediately before or after daily F 23
treatment, IGRT ensures patient positioning accuracy and delivery Age (y)
verification for each radiation treatment (6, 7). Range 22-93
We compared daily interfraction variations in setup for patients Mean 58.8
with head-and-neck cancer from 3 different IGRT modalities. The Median 60.0
daily shifts from 3 IGRT techniques, including 632 KVCBCT, 974 Initial weight (lb)
MVFBCT, and 1696 MVCBCT scans, in mediolateral (ML), Range 87.4-278.6
craniocaudal (CC), and anteroposterior (AP) directions, were Mean 169.0
retrospectively analyzed. The purposes of this work were (1) to Median 160.6
explore how different IGRT techniques affect patient setup error Weight change (lb)
accuracy, (2) to estimate clinical target volume-to-planned target Range 1.8-33.8
volume (CTV-to-PTV) margins based on the daily shifts, and (3) Mean 13.4
to investigate possible correlations between patient/tumor char- Median 12.8
acteristics and interfraction variations in patient setup. To our Weight change (%)
knowledge, this is the first study evaluating these effects in the Range 1.6-22.7
H&N patient population. Mean 7.9
BMI (kg/m2)
Methods and Materials <18.5 2
18.5-24.9 15
24.9-29.9 17
Study description
>30.0 7
Unknown 12
A total of 3302 CT images for 117 patients with head-and-neck Tumor location
cancer who underwent IMRT (2007-2009) with daily IGRT were Upper neck 13
analyzed. Similar imaging quality assurance (QA) procedures and Middle neck 27
daily IGRT protocols were used for all patients from 3 partici- Lower neck 11
pating institutes. For each patient, daily CT scanning was per- Unknown 2
formed before each treatment using KVCBCT, MVFBCT, or
Abbreviation: BMI Z body mass index; MVCBCT Z megavoltage
MVCBCT as image guidance. We studied 29 patients treated on an
cone-beam computed tomography.
Elekta, 35 patients on a TomoTherapy, and 53 patients on
434 Qi et al.
demographics are shown in Table 1. To analyze various correlations,
we further divided the group into subgroups by tumor location and
other patient characteristics. The cutoff points for body mass index
(BMI, kg/m2) were underweight (BMI<18.5), normal
(18.5BMI25), and overweight (BMI>25). For different subsites
of head-and-neck cancer, the upper (nasopharynx), middle (oral
cavity), and lower parts of the neck were considered. The analysis of
variance (ANOVA) model and/or Student t test were used to test
whether there was a significant correlation between patient setup
and patient characteristics.
CT-based IGRT systems
Elekta Synergy-S (Elekta Inc, Norcross, GA) is equipped with
a kV(100-130 kVp) cone-beam CT (KVCBCT) with an x-ray based
volumetric imaging system for patient setup. The kV x-ray source
and amorphous silicon imaging panel are mounted opposing each
other and orthogonal to the treatment beam. Interfractional patient
displacements can be quantified and adjusted daily using KVCBCT
before treatment.
Siemens Avant-Garde (System Healthcare, Concord, CA), on
the other hand, uses megavoltage cone-beam CT (MVCBCT) with
6MV photons by rotating the gantry continuously from 270
through 360 to 110 at a source-to-image distance of 145 cm.
TomoTherapy (Accuray Inc, Sunnyvale, CA) enables a helical
fan beam scan using detuned 3.5MV photons and an arc-shaped
xenon CT detector array mounted on the opposite side of the ring
gantry. The source-to-detector distance is 145 cm. The imaging
field of view of 40 cm is defined by the width of the TomoTherapy
multileaf collimator. The utility of an on-board MVCT fan beam
(MVFBCT) system on the TomoTherapy unit allows the align-
ment of the patient’s anatomy using a relatively low dose per
image acquisition.
IGRT procedure
The utility of IGRT allows for the verification of patient’s anatomy
and alignment through relatively low-dose CT imaging. Inter-
fraction patient setup is analyzed and adjusted so that daily radi-
ation treatments can be accurately delivered. The routine IGRT
procedure includes setting up the patient using in-room lasers and
individual skin markers/tattoos, acquisition of pretreatment CT
images, image registration, and visual inspection, and if necessary
making the appropriate shift.
Initial patient setup
A standard head holder and a customized head-and-neck and
shoulder thermoplastic mask (MEDTEC S-frame) were used for
all patients to provide immobilization of the entire upper part of
the body in the treatment position. The mask was fabricated
during the initial CT simulation. Before each treatment, alignment
was performed using in-room lasers and 3-point markers on the
patient with the customized immobilization device.
Image acquisition
The daily image scan was acquired for each patient in the treat-
ment position immediately before each radiation delivery. For the
Elekta KVCBCT, the daily image acquisition of a 360 cone-beam
International Journal of Radiation Oncology  Biology  Physics
CT scan and image reconstruction usually took less than 2
minutes. For the Siemens MVCBCT, the standard head-and-neck
cancer protocol of 5 monitor units was used for each scan, with an
acquisition time of 2 minutes. TomoTherapy image acquisition
usually required 2 to 5 minutes, using a normal slice thickness of 4
mm for the daily MVFBCT.
Image registration and IGRT protocol
The pretreatment scan was registered to the reference image set
(the CT simulation image) to verify patient positioning or to
determine whether repositioning was required. For all the exam-
ined patients, the initial alignment was automated using bony
structures, followed by anatomy-based manual registration if
necessary.
The initial registration for the 3 image modalities was based on
bony structures. Specifically, the C2 vertebra, the C6 vertebra, or
both were identified for registration of the region of interest. On
TomoTherapy, this was conducted by aligning the pretreatment
MVFBCT scan to the reference CT images using automatic
registration that used a mutual information algorithm (8). For the
Elekta KVCBCT, the coarse adjustment was done using the
automated bone-matching mode (9) within the defined image
guidance volume. The Siemens MVCBCT used an automated
maximization of mutual information algorithm (10). Each of these
3 IGRT systems provided 3-dimensional image registration. If
visual inspection in the axial, sagittal, and coronal views indicated
that further adjustment was necessary, manual fine tuning was
subsequently used.
The initial registration and adjustment were done by the
therapists, and the fine tuning was always performed manually by
comparing all relevant anatomic landmarks. When relatively large
adjustments were required (ie, >1 cm), the physician was called to
verify the alignment on the treatment console.
Image quality verification and quality assurance
All machines used in this study at 3 institutions were credentialed
by the Radiological Physics Center. QA of all 3 IGRT image
systems was performed monthly to ensure image quality in
specification (11). In addition, positioning reproducibility was
tested daily for MVCBCT and KVCBCT (using cubic phantoms)
and MVFBCT (using a cheese phantom, which is a 2-
hemicylinder solid water phantom supplied by TomoTherapy to
perform various quality assurance procedures). Initial IGRT scans
were performed daily to generate table offsets by positioning the
phantom at known shifts. A second scan was generally performed
to estimate the residual positioning error for a rigid body system.
The mean residual positioning errors of 0.5 to 1.0 mm were
observed in 3 translational directions.
Systematic error, random error, and CTV-to-PTV
margin
Systematic errors (S) represent displacements that are consistent
during the course of treatment; and random error (s) represents
day-to-day variations in the patient setup. S is calculated as the
standard deviation of the average setup deviations per patient in
the group of patients, and s is defined as the root mean square of
the standard deviation of all patients. The CTV-to-PTV margin is
Volume 86  Number 3  2013 Interfraction patient setup for head-and-neck IMRT 435

calculated using the equation 2.5 S þ 0.7 s to ensure a minimum
dose of 95% prescription dose to cover the CTV for 90% of the
patient population (12). The estimated margin, based on the data
derived from IGRT, applies to patients treated on the machines
where no image guidance capability is available. A uniform
margin, defined as the maximal calculated margin among the 3
translational directions, was reported.
Data/statistical analysis
Analysis of variance (ANOVA) and/or, Student t test, were used to
explore how image modalities and patient characteristics affected
patient repositioning in the ML, CC, and AP directions. A P value
of .05 or smaller was considered to be statistically significant.

Interobserver variations
The interobserver variations were evaluated. Twenty users,
including 3 physicians, 3 physicists, 4 dosimetrists, and 10
therapists, participated in the study. A representative daily scan
image for each modality was given to the participants. To
simulate the realistic IGRT workflow, automatic registration was
done before manual registration. Shifts were then recorded and
analyzed.
Results
Figure 1 shows the average shifts and standard deviations in the
ML, CC, and AP directions using the evaluated systems. Patients 1
through 29 were scanned with KVCBCT, patients 30 through 65
with MVFBCT, and patients 66 through 117 with MVCBCT.
Table 2 shows the average interfraction shifts and standard
deviations in the ML, CC, and AP directions using 3 image
modalities. One-way blocked ANOVA analysis found no statistical
significance in the average shifts across all 3 imaging modalities

Fig. 1. Average shifts and standard deviations in the mediolateral (ML), craniocaudal (CC), and anteroposterior (AP) directions using
kilovoltage cone-beam computed tomography (KVCBCT), megavoltage fan-beam computed tomography (MVFBCT), and megavoltage
cone-beam computed tomography (MVCBCT) for patients with head-and-neck cancer. Patients 1 through 29 were scanned with KVCBCT,
patients 30 through 64 were scanned with MVFBCT, and patients 65 through 117 were scanned with MVCBCT.
436 Qi et al. International Journal of Radiation Oncology  Biology  Physics

Table 2 Average daily shifts and standard deviations for 117 patients with head-and-neck cancer imaged with KVCBCT, MVFBCT,
and MVCBCT
KVCBCT MVFBCT MVCBCT
Direction Mean (mm) SD (mm) Mean (mm) SD (mm) Mean (mm) SD (mm)
ML 0.5 1.5 0.2 1.9 0.0 1.8
CC 0.3 2.0 0.2 2.4 0.5 1.7
AP 0.3 1.7 0.0 1.7 0.8 3.1
Abbreviations: AP Z anteroposterior; CC Z craniocaudal; KVCBCT Z kilovoltage cone-beam computed tomography; ML Z mediolateral;
MVCBCT Z megavoltage cone-beam computed tomography; MVFBCT Z megavoltage fan-beam computed tomography; SD Z standard deviation.

with P values of .50 (ML), .09 (CC), and .11 (AP) respectively.
KVCBCT had the smallest standard deviations in the ML direction
and equal or smaller standard deviations in the AP direction,
compared with the 2 megavoltage-based image modalities, whereas
MVCBCT yielded the largest AP variations.
Distributions of interfraction shifts in the ML, CC, and AP
directions for 117 patients with head-and-neck cancer imaged with
the 3 evaluated image modalities are shown in Figure 2. Averaged
across the entire treatment course, 41.5%, 32.1%, and 52.8% of
the MVCBCT patients had displacements 3 mm in the ML, CC,
and AP directions, respectively, whereas for KVCBCT and
MVFBCT, the displacements 3 mm in these 3 directions were
10.3%, 31.0%, 20.7%, and 17.1%, 25.7%, and 20.0%, respec-
tively. Displacements 5 mm were seen only in the CC direction
for both KVCBCT and MVFBCT, whereas in MVCBCT, they
were seen in all 3 directions, ranging from 1.9% (ML and CC) to
17% (AP).
Systematic error (S), random error (s), and calculated CTV-to-
PTV margins in the ML, CC, and AP directions for the 3 IGRT
systems are shown in Table 3. The calculated margins were found
to be anisotropic in 3 translational directions. The rotational
deviations were ignored in the margin calculation because of the
unavailability of such corrections in the KVCBCT and MVCBCT
systems. We observed the largest random setup errors in
MVCBCT (range, 2.0-2.1 mm) in all 3 translational directions.
Random errors were the smallest in KVCBCT (range, 1.4-1.6
mm), compared with MVFBCT (1.7 mm) and MVCBCT. The
calculated margins were found to be anisotropic in the trans-
lational directions for the 3 considered image modalities (ie, the
maximum margins for KVCBCT were 5.6 mm in the CC
direction, whereas the maximum margins for MVFBCT and
MVCBCT were 6.9 mm (CC) and 8.9 mm (AP), respectively). If
a uniform margin was required, it would be 5.6 mm, 6.9 mm, and
8.9 mm for KVCBCT, MVFBCT, and MVCBCT, respectively,
when IGRT is not present.
The correlations between CTV-to-PTV margins and various
patient characteristics and tumor locations were further investi-
gated based on MVCBCT data. Table 4 summarizes the S, s, and
calculated CTV-to-PTV margins in the ML, CC, and AP directions
and the P values for the statistical differences between the
subgroups. Although the calculated margins for different patient
parameters appeared to differ, these differences were generally not
statistically significant.
The standard deviations of interobserver rigid registration
variations were found to be 0.8, 1.1, and 0.7 mm (MVCBCT); 0.5,
0.4, and 0.8 mm (MVFBCT); and 0.5, 0.4, and 0.6 mm
(KVCBCT) in the ML, CC, and AP directions, respectively. All
variations were small (w1 mm), and no statistical differences
were found between modalities.
Discussion
The advantage of IGRT is to provide a more accurate and repro-
ducible patient setup to achieve more accurate dose delivery that
may translate into better tumor control and reduction of treatment-
related toxicity. However, the daily shifts and margins recom-
mended by various IGRT technologies may vary, depending on
image quality, image registration methods, and other imaging
parameters. Image quality is influenced by photon energies used

Fig. 2. Distribution of interfraction displacements in the (a) mediolateral (ML), (b) craniocaudal (CC), and (c) anteroposterior (AP)
directions for intensity modulated radiation therapy of head-and-neck cancer using Elekta kilovoltage cone-beam computed tomography
(KVCBCT), TomoTherapy megavoltage fan-beam computed tomography (MVFBCT), and Siemens megavoltage cone-geam computed
tomography (MVCBCT).
Volume 86  Number 3  2013 Interfraction patient setup for head-and-neck IMRT 437

Table 3 Systematic errors, random errors, and calculated CTV-to-PTV margins in mediolateral, craniocaudal, and anteroposterior
directions
KVCBCT MVFBCT MVCBCT
Direction S (mm) s (mm) Margin* (mm) S (mm) s (mm) Margin* (mm) S (mm) s (mm) Margin* (mm)
ML 1.0 1.4 3.6 1.2 1.7 4.1 1.9 2.0 6.1
CC 1.8 1.6 5.6 2.3 1.7 6.9 1.8 2.1 5.9
AP 1.2 1.4 4.0 1.1 1.7 3.9 3.0 2.0 8.9
Abbreviations: AP Z anteroposterior; CC Z craniocaudal; CTV Z clinical target volume; KVCBCT Z kilovoltage cone-beam computed tomog-
raphy; ML Z mediolateral; MVCBCT Z megavoltage cone-beam computed tomography; MVFBCT Z megavoltage fan-beam computed tomography;
PTV Z planned target volume; s Z random errors; S Z systematic errors.
* Rotational deviations were ignored in the margin calculation.

for imaging and geometries of source and detectors, among many
other factors. Owing to the use of megavoltage x-ray beams and
the relatively large size of detectors in the image panel for
TomoTherapy MVFBCT and Siemens MVCBCT images,
KVCBCT demonstrates superior spatial resolution and contrast
(13). The TomoTherapy MVFBCT, with the same geometric
design as a diagnostic CT, has a lower contrast-to-noise ratio
because of a megavoltage photon beam for imaging. The Siemens
MVision, using MVCBCT with 6-MV photons and a flat-panel
imager, yields the lowest image quality among the 3 image
systems considered in this study (14). Based on phantom studies
performed by other researchers (13, 14), higher spatial resolution
was found in KVCBCT than in MVFBCT and MVCBCT, and
MVCBCT had the poorest sensitometry and uniformity among the
3. The imaging quality appeared to be in agreement with the
random errors shown in Table 3. We further analyzed other factors

Table 4 Systematic errors, random errors, and calculated CTV-to-PTV margins in mediolateral, craniocaudal, and anteroposterior
directions for different patient characteristics and tumor locations
S s Margin
ML (mm) CC (mm) AP (mm) ML (mm) CC (mm) AP (mm) ML (mm) CC (mm) AP (mm)
Entire cohort 1.9 1.8 3.0 2.0 2.1 2.0 6.1 5.9 8.9
Sex
F 1.5 1.8 2.9 1.9 1.8 2.3 5.2 5.7 8.6
M 2.1 1.7 3.1 2.1 1.8 2.0 6.8 5.5 9.2
P values .68 .12 .25 .60 .24 .81
Age (y)
<60 1.5 1.7 2.2 1.8 2.0 1.9 5.0 5.7 6.8
>60 2.1 1.9 3.7 2.2 2.3 2.0 6.8 6.4 10.7
P values .09 .82 .73 .07 .15 .42
Weight (lb)
<169 1.5 1.7 2.5 1.8 1.6 2.2 5.2 5.4 7.8
>169 2.1 1.8 3.3 2.3 2.0 2.0 6.9 5.9 9.7
P values .58 .05 .51 .22 .32 .81
Weight loss (lb)
<15 2.0 1.8 3.0 2.1 1.8 2.2 6.5 5.8 9.0
>15 1.7 1.7 3.0 2.0 1.7 2.0 5.7 5.4 8.9
P values .06 .31 .78 .72 .39 .34
% Weight loss
<7.9% 2.1 1.9 2.9 2.1 2.2 1.9 6.7 6.3 8.6
>7.9% 1.7 1.5 2.9 2.0 2.1 2.1 5.7 5.2 8.7
P values .55 .64 .08 .60 .98 .34
BMI (kg/m2)
24.9 1.9 1.3 2.7 2.1 2.0 2.1 6.2 4.7 8.2
>24.9 2.3 1.9 3.2 2.2 2.0 2.3 7.3 6.2 9.6
P values .99 .79 .57 .96 .42 .19
Subsites
Upper 1.7 2.1 2.6 1.6 1.9 1.9 5.3 6.7 10.4
Middle 1.5 1.5 2.7 1.9 2.1 1.8 5.0 5.3 8.1
Lower 2.8 1.5 2.8 2.5 2.3 2.0 8.7 5.4 8.5
P values .83 .12 .42 .07 .09 .56
Abbreviations: AP Z anteroposterior; BMI Z body mass index; CC Z craniocaudal; CTV Z clinical target volume; ML Z mediolateral;
PTV Z planned target volume; s Z random errors; S Z systematic errors.
438 Qi et al.
that might affect the setup errors and the CTV-to-PTV margins in
the MVCBCT group; no statistical differences were found
between the sex, age, weight, and tumor location subgroups. Our
result is in agreement with that of Morrow et al (15), in which
superior image quality resulted in better reproducibility of soft
tissue-based registration for IGRT of the prostate.
The margin calculation is a quantitative measure to illustrate
the difference for each IGRT cohort. Many factors, such as IGRT
procedures, immobilization devices, and operator expertise level,
can affect the magnitude of systematic errors and random errors,
and therefore affect the CTV-to-PTV margin. In this study, we
found that image modality further complicates the margin for
IMRT for head-and-neck cancer, where bony structure was
generally used for image registration. Superior imaging quality
determined from KVCBCT provided the smallest CTV-to-PTV
margin among the 3 IGRT modalities investigated. To avoid
potentially large uncertainties caused by poor imaging quality for
margin estimation, we encourage that margins be estimated from
KVCBCT when IGRT is not present. When IGRT is available, the
average CTV-to-PTV margins can be estimated according to Eq.
(3) in Den et al (7). Based on the current data, non-patient
dependent maximal margins of 3.1, 3.3, and 4.2 mm were needed
for KVCBCT, MVFBCT, and MVCBCT cohort respectively. A
limitation of this work is that we considered only rigid image
registration. Residual uncertainties, such as patient intrafractional
motion and anatomic changes, need to be accounted for in
determining the CTV-to-PTV margins if nonrigid image regis-
tration is considered for IMRT of head-and-neck cancer (16).
Other than the image quality and patient characteristics,
factors such as image registration algorithms and image acquisi-
tion parameters used in IGRT protocols may also affect setup
errors and margins for different image modalities. The magnitude
of the shifts attributed to image acquisition and automatic regis-
tration, however, is difficult to assess, because each specific
algorithm is an integral part of the imaging system of a specific
vendor. For example, larger slice spacing of 4 mm (normal
acquisition mode) for MVFBCT, compared with the reconstruc-
tion slice spacing of 1 mm for MVCBCT and KVCBCT, resulted
in larger variability in translational shifts in the CC direction as
well as in other directions on TomoTherapy (17). The potential
table sagging (1-2 mm) might account for the largest systematic
error in the AP direction for Siemens MVCBCT. The registration
variations may also be influenced by the sizes, shapes, and
positions of the region of interest, according to Cui et al (18).
Furthermore, the head-and-neck region is not a rigid body, and the
alignment and residual errors will depend on the landmarks used
for registration (19).
The variations of interobserver manual registration showed
minimal difference between different image modalities; larger
variations appeared to be associated more with MV image guid-
ance than with KVCBCT. Although similar IGRT protocols,
landmarks, and QA criteria were used, the images were acquired
at 3 different institutions with different therapists of different
levels of expertise, which could confound the findings of the study
(ie, 2 or 3 therapists were involved in imaging acquisition/regis-
tration process for KVCBCT and MVFBCT, and the entire ther-
apist team (w8 therapists) was rotated on the Siemens unit).
However, the patient population and number of images are suffi-
ciently large to warrant the conclusions.
The imaging doses remain a concern for nearby critical
structures in IGRT procedures. Additional doses per scan of 0.3 to
4 cGy for Elekta KVCBCT, 1 to 3 cGy for TomoTherapy, and 2 to
International Journal of Radiation Oncology  Biology  Physics
16 cGy for Siemens MVCBCT have been reported (13, 14, 20).
This extra imaging dose should be reduced to as low a reasonable
level as possible or taken into account in the process of treatment
planning. For better reproducibility and lower imaging dose, the
IGRT systems with better image quality appear to be a better
choice if they are available.
Conclusion
Random day-to-day setup errors for IGRT and CTV-to-PTV
margins for non-IGRT of head-and-neck cancer may be affected
by the imaging modality used to assess the setup error. Our data
indicate that larger margins were associated with MVFBCT and
MVCBCT, compared with a smaller margin for KVCBCT.
IGRT modalities with better image quality are encouraged in
clinical practice.
References
1. Butler EB, Teh BS, Grant WH 3rd, et al. SMART (simultaneous
modulated accelerated radiation therapy) boost: A new accelerated
fractionation schedule for the treatment of head and neck cancer with
intensity modulated radiotherapy. Int J Radiat Oncol Biol Phys 1999;
45:21-32.
2. Manning MA, Wu Q, Cardinale RM, et al. The effect of setup
uncertainty on normal tissue sparing with IMRT for head-and-neck
cancer. Int J Radiat Oncol Biol Phys 2001;51:1400-1409.
3. Jaffray DA, Siewerdsen JH. Cone beam computed tomography with
a flat-panel imager: Initial performance characterization. Med Phys
2000;27:1311-1323.
4. Mackie TR, Holmes T, Swerdloff S, et al. Tomotherapy: A new
concept for the delivery of dynamic conformal radiotherapy. Med Phys
1993;20:1709-1719.
5. Pouliot J, Bani-Hashemi A, Chen J, et al. Low-dose megavoltage
cone-beam CT for radiation therapy. Int J Radiat Oncol Biol Phys
2005;61:552-560.
6. Zumsteg Z, DeMarco J, Lee SP, et al. Image guidance during head-
and-neck cancer radiation therapy: Analysis of alignment trends
with in-room cone-beam computed tomography scans. Int J Radiat
Oncol Biol Phys 2012;83:712-719.
7. Den RB, Doemer A, Kubicek G, et al. Daily image guidance with
cone-beam computed tomotherapy for head-and-neck cancer intensity-
modulated radiotherapy: A prospective study. Int J Radiat Oncol Biol
Phys 2010;76:1353-1359.
8. Ruchala K, Olivera GH, Kapatoes J. Limited-data image registration
for radiotherapy positioning and verification. Int J Radiat Oncol Biol
Phys 2002;54:592-605.
9. Borgefors G. Hierarchical chamfer matching: A parameter edge
matching algorithm. IEEE Trans Pattern Anal Mach Intell 1988;10:
849-865.
10. Pluim JP, Maintz JBA, Viergever MA. Mutual-information-based
registration of medical images: A survey. IEEE Trans Med Imaging
2003;22:986-1004.
11. Bissonnette JP, Balter PA, Dong L, et al. Quality assurance for image-
guided radiation therapy utilizing CT-based technologies: A report of
the AAPM TG-179. Med Phys 2012;39:1946-1963.
12. Van Herk M. Errors and margins in radiotherapy. Semin Radiat Oncol
2004;14:52-64.
13. Chan M, Yang J, Song Y, et al. Evaluation of imaging perfor-
mance of major image guidance systems. Biomed Imaging Interv J
2011;7:1-7.
14. Stutzel J, Oelfke U, Nill S. A quantitative image quality comparison of
four different image guided radiotherapy devices. Radiother Oncol
2008;86:20-24.
Volume 86  Number 3  2013
15. Morrow NV, Lawton CA, Qi XS, et al. Impact of computed tomog-
raphy image quality on image-guided radiation therapy based on soft
tissue registration. Int J Radiat Oncol Biol Phys 2012;82:e733-e738.
16. Yang JZ, Garden AS, Zhang YB, et al. Variable planning margin
approach to account for locoregional variations in setup uncertainties.
Med Phys 2012;39:5136-5144.
17. Woodford C, Yartsev S, van Dyk J. Optimization of megavoltage CT
scan registration settings for brain cancer treatments on tomotherapy.
Phys Med Biol 2007;52:185-193.
Interfraction patient setup for head-and-neck IMRT 439
18. Cui YF, Galvin JM, Straube WL, et al. Multi-system verification of
registrations for image-guided radiotherapy in clinical trials. Int J
Radiat Oncol Biol Phys 2011;81:305-312.
19. Lee C, Langen KM, Lu W, et al. Assessment of parotid gland dose
changes during head and neck cancer radiotherapy using daily meg-
avoltage computed tomography and deformable image registration. Int
J Radiat Oncol Biol Phys 2008;71:1563-1571.
20. Amer A, Marchant T, Sykes J, et al. Imaging doses from the Elekta
Synergy X-ray cone beam CT system. Br J Radiol 2007;80:476-482.