validation of a new formulism and the related correction ... · pdf filedifficulties in...
TRANSCRIPT
Validation of a new formulism and the related correction factors on output factor
determination for small photon fields Yizhen Wang
*1, Kelly Younge
1, Michelle Nielsen
1, Theodore Mutanga
1, Congwu Cui
1,2, Indra J. Das
3
(1)Peel Regional Cancer Center, Trillium Health Partners, Mississauga, ON (2) Department of Radiation
Oncology, University of Toronto, Toronto, ON (3) Radiation Oncology Dept., Indiana University- School of
Medicine, Indianapolis, IN, USA
Introduction
It is well known that the small field sizes (≤30 x 30mm2) encountered in stereotactic radiosurgery (SRS) create
difficulties in measuring dosimetric data that arise due to lack of charged-particle equilibrium or transient
equilibrium (CPE/TCPE) and occlusion of the radiation source.1, 2
. Furthermore, the choice of detector used for
measurements in small field conditions is complicated by the finite size of the detector and non-water equivalence of
detector components.3 Significant variations in measured output factors or total scatter factors using different
detectors in the same measurement conditions have been reported.1, 4-6
If uncorrected, such variations could
potentially lead to significant error in delivered dose.7
Micro-ionization chambers such as the Exradin A16 chamber (Standard Imaging, Middleton, WI, USA) with 0.007
cm3 volume have been suggested for total scatter factor measurements in small fields.
3 However, due to the presence
of non-water equivalent chamber components (wall, central electrode, and stem) as well as volume effects that
become pronounced in regions of electronic disequilibrium, corrections are needed to account for the perturbation of
particle fluence.
Diode detectors such as the EDGE (Sun Nuclear, Melbourne, FL, USA) and SFD (IBA, Schwarzenbruck, Germany)
detectors, have been widely used for small field output factor measurements.1 The advantages of diodes in terms of
small size and high SNR are well understood, however in addition to energy dependence of these detectors, typical
construction includes high Z materials (brass, copper) in detector packaging leading to significant fluence
perturbations.8 Radiochromic film has also been used for small field commissioning
9 with its main advantages being
water equivalence and very high spatial resolution. However, variations in measured dosimetric data with film have
been attributed to user handling (scanning, post irradiation signal loss, etc) as well as non-uniformity in film
designs.3
A new formulism has been suggested to correct the detector response in small photon fields.10
The correction factors
used in the formulism have been determined for various detectors.1, 11, 12
In the current study, we present results of
our experience with the measurement of output factors using three different detectors, EDGE, SFD and A16, during
the commissioning of our linac-based stereotactic radiosurgery program on a Varian Clinac iX linear accelerator
(Varian Medical System, Palo Alto, CA) with BrainLab conical collimators (BrainLab, Westchester, IL). The
purpose of the publication of these results is twofold: 1) to provide small field dosimetric data that are relatively
scarce in the literature and 2) to serve as a validation of the published correction factors for various detectors used in
small field dosimetric data measurement.
Methods
A. Correction Factor for Small Field Dosimetry
As per the new formulism the absorbed dose to water at a point in a phantom for a clinical field, fclin, of quality Qclin
in the absence of the dosimeter is given by:
msrclin
msrclin
msr
msr
clin
clin
ff
f
Qw
f
Qw DD,
,,, (1)
The factor msrclin
msrclin
ff
,
, converts absorbed dose to water for the machine-specific reference field (msr) to the absorbed
dose to water for the clinical field. msrclin
msrclin
ff
,
, can be expressed as a ratio of detector readings multiplied by a
detector response correction factor msrclin
msrclin
ff
QQk,
, which can be derived from Monte Carlo simulation or obtained from
measurement, i.e.,
msrclin
msrclinmsr
msr
clin
clinmsrclin
msrclin
ff
QQf
Q
f
Qff
QQ kM
M,
,
,
, (2)
Therefore, for small fields the total scatter factor is no longer equal to the ratio of detector readings. Thus, msrclin
msrclin
ff
QQk,
, needs to be applied to correct the detector’s response for small fields. Table 1 lists the k correction factors
used in this study. Note that the k factors for the EDGE detector and the A16 chamber were derived for Siemens and
Elekta machines11
and those for the SFD were derived for Varian machines12
. The original k factors published do
not include all the field sizes used in this study, therefore interpolation was performed to obtain the k factors for
some field sizes.
Table 1. msrclin
msrclin
f,f
Q,Qk factors used in this study
Cone size, diameter (mm) 5 7.5 10 12.5 15 17.5 20 25 30
EDGE11
0.932 0.951 0.967 0.978 0.986 0.989 0.991 0.996 1.001
A1611
1.112 1.044 1.020 1.007 1.002 1.001 1.001 1.000 0.999
SFD12
0.964 0.982 0.992 0.994 0.996 0.998 1.000 1.003 1.005
B. Detectors and Devices
The Exradin A16 micro chamber has a collecting volume of 0.007 cm3, 1.7 mm outer shell collecting volume radius,
1.2 mm inner collecting volume radius, and 2.4 mm collecting volume length. It is constructed using air-equivalent
plastic (C552) with a central electrode of silver-plated copper covered steel. The EDGE detector has an active area
of 0.8x0.8 mm2 and a thickness of 0.03 mm packaged in brass shielding. The SFD has an active area of 0.6 mm
diameter and a thickness of 0.06 mm packaged in plastic ABS and epoxy resin. The EDGE is a shielded detector
while the SFD is an unshielded detector.
C. Measurement Methods
Brainlab conical cones of 5, 7.5, 10, 12.5, 15, 17.5, 20, 25, 30 mm diameters are employed to collimate a 6 MV
photon beam on a Varian Clinac iX linear accelerator for the stereotactic radiosurgery treatment at our institution.
Measurements were performed using a scanning water phantom. Following the requirements of our planning
system, the dose was measured at a depth of 1.5 cm with 98.5 cm SSD to the water surface. Beam profile scans
along both transverse and radial directions were performed to ensure the centering of each detector prior to the
output factor measurements. All output factors are normalized to a 10 x 10 cm2 field, while a cross calibration
technique (daisy chain method) was adopted for diode measurement results to account for differences of diode
detector response between small and large photon fields. The intermediate field was the 3 cm cone field.
Results and Discussion
The original output factors measured before correction are shown in Table 2 and Fig. 1. Large discrepancies up to
20% were observed for small fields, i.e., 5, 7.5 and 10 mm diameter cones. For fields larger than 10 mm, the
discrepancies are less than 4%.
Table 2. Output factor measured before correction. All results are normalized to 10x10 cm2 field.
Cone size, diameter (mm) 5 7.5 10 12.5 15 17.5 20 25 30
EDGE 0.694 0.795 0.846 0.875 0.890 0.900 0.905 0.910 0.914
A16 0.577 0.720 0.797 0.843 0.870 0.889 0.899 0.909 0.915
SFD 0.675 0.763 0.818 0.851 0.871 0.884 0.893 0.901 0.909
The correction factors in Table 1 were applied to the output factors in Table 2. The results are shown in Table 3 and
Fig. 2. After correction the discrepancies among various detectors are ~1% or smaller. The average of the corrected
output factors are adopted for our SRS planning system. It can be seen that without correction, diodes overestimate
the output factor for very small fields (especially the shielded EDGE detector). The A-16 ion chamber, one of the
smallest ion chambers on the market, underestimates the output factors. Consistent output factors among the three
detectors were obtained based on the correction factors in Table 1 though the correction factors were determined on
various linacs.
Conclusions
Caution is needed when determining the output factors for small photon fields, especially for the fields 10 mm in
diameter or smaller. More than one type of detector should be used, each with proper corrections applied to the
measurement results. It is concluded that with the application of correction factors to appropriately chosen detectors,
output can be measured accurately for small fields.
Fig. 1. Output factor measured before correction.
Fig. 2. Output factor measured after correction.
Table 3. Output factor after correction applied. All results are normalized to a 10x10 cm2 field.
Cone size, diameter (mm) 5 7.5 10 12.5 15 17.5 20 25 30
EDGE corrected 0.647 0.756 0.818 0.856 0.878 0.890 0.897 0.907 0.915
A16 corrected 0.642 0.751 0.813 0.849 0.872 0.890 0.900 0.909 0.914
SFD corrected 0.650 0.749 0.811 0.846 0.868 0.882 0.893 0.904 0.914
References: 1 C. Bassinet et al., “Small fields output factors measurements and correction factors determination for several
detectors for a CyberKnife(®) and linear accelerators equipped with microMLC and circular cones.,” Med. Phys.
40(7), 071725 (2013). 2 O.A. Sauer, “Determination of the quality index (Q) for photon beams at arbitrary field sizes,” Med Phys 36(9),
4168–4172 (2009). 3 I.J. Das, M.B. Downes, A. Kassaee, and Z. Tochner, “Choice of Radiation Detector in Dosimetry of Stereotactic
Radiosurgery-Radiotherapy,” J. Radiosurgery 3(4), 177–186 (n.d.). 4 P. Francescon, S. Cora, and C. Cavedon, “Total scatter factors of small beams: A multidetector and Monte Carlo
study,” Med. Phys. 35(2), 504 (2008). 5 J.P. Manens, I. Buchheit, H. Beauvais, G. Gaboriaud, A. Mazal, and P. Piret, “Dosimetry of small-size photon
beams,” Cancer Radiother 2(2), 105–114 (1998). 6 G. Khelashvili, J. Chu, A. Diaz, and J. Turian, “Dosimetric characteristics of the small diameter brainLAB
TM
cones used for stereotactic radiosurgery,” J. Appl. Clin. Med. Phys. 13(1), 4–13 (2012). 7 D.S. Followill et al., “The Radiological Physics Center ’ s standard dataset for small field size output factors,”
13(5), 282–289 (2012). 8 H.-J.J. Shin et al., “Evaluation of the EDGE detector in small-field dosimetry,” J. Korean Phys. Soc. 63(1), 128–
134 (2013). 9 J.M. Larraga-Gutierrez, D. Garcia-Hernandez, O.A. Garcia-Garduno, O.O. Galvan de la Cruz, P. Ballesteros-
Zebadua, and K.P. Esparza-Moreno, “Evaluation of the Gafchromic EBT2 film for the dosimetry of radiosurgical
beams,” Med Phys 39(10), 6111–6117 (2012). 10
P. Alfonso, P. Andreo, R. Capote, M. S. Huq, W. Kilby, P. Kjall, T. R. Mackie, H. Palmans, K. Rosser,
J.Seuntjens, W. Ullrich and S. Vatnitsky, "A new formalism for refeence dosimetry of small and nonstandard
fields," Med Phys 35, 5179-5186 (2008). 11
P. Francescon, S. Cora and N. Satariano, "Calculation of
k(Q(clin),Q(msr) ) (f(clin),f(msr) ) for several small detectors and for two linear accelerators using Monte Carlo
simulations," Med Phys 38, 6513-6527 (2011) 12
G. Cranmer-Sargison, S. Weston, J. A. Evans, N. P. Sidhu and D. I. Thwaites, "Implementing a newly proposed
Monte Carlo based small field dosimetry formalism for a comprehensive set of diode detectors," Med Phys 38,
6592-6602 (2011).
A comparison of dose reduction methods on image quality for cone beam CT
R Webb1,2, LA Buckley*1,
(1) The Ottawa Hospital Cancer Centre (2) presently at Elekta Inc
Introduction
Modern radiotherapy techniques make use of highly conformal dose distributions. This complexity limits the dose to
normal tissues and permits the use of dose escalation in cases where it is clinically advantageous. A high degree of
conformality is only useful if the patient position can be known with a high level of precision and accuracy. More
stringent demands on the accuracy of the patient positioning have led to increased use of sophisticated image
guidance techniques.
Kilovoltage cone beam computed tomography (CBCT) is one such technique that is routinely used for positioning
verification. The CBCT x-ray system is mounted to the gantry and is typically used daily to acquire an image of the
target volumes with the patient in the treatment position prior to treatment. A consequence of this technique is the
high patient dose received from the kV-CBCT imaging relative to planar imaging techniques. While the dose from
imaging is small relative to the treatment dose, routine use of these techniques can lead to significant dose to the
normal tissues. Several studies have investigated the dose from kV-CBCT and have proposed methods to reduce the
dose. It is however essential that image quality not be compromised such that it affects the accuracy of the image
registration. This study evaluates the impact of a variety of imaging parameters on both the dose and the image
quality of a clinical kV-CBCT system. It investigates how changes to the CBCT image settings can be used to
reduce the patient dose while maintaining a comparable level of image quality.
Methods
All dose and image quality measurements were performed on an Elekta Synergy S linac (Elekta, Norcross GA) with
XVI cone-beam CT. All dose measurements were taken on a single machine. The image quality measurements were
performed on multiple XVI units using software versions 4.2.1 and 4.5. Flexmap, image offset and multi-level gain
calibrations were performed before all measurements and collimation and filtration was unchanged across all
measurements.
Dose measurement methodology was based on AAPM report 111 and used an NE2571 Farmer chamber in a 32cm x
45cm phantom consisting of three CTDI body phantoms place end-to-end. Doses were computed using the standard
cone-beam dose index (CBDI) weighting (1/3 central dose + 2/3 peripheral dose) and with the alternative weighing
proposed by Bakalyar. Image quality was assessed using a Catphan CTP500 phantom at isocentre and was based on
the XVI4.5 customer acceptance test procedure.
Imaging presets were created for a range of exposure conditions based on the standard chest preset (120kV, 25mA,
40ms). Tube voltage was varied from 70 to 140kV in 8 steps, tube current was varied from 10 to 100mA in 11 steps
and the number of projections was varied from 330 to 1243 frames in 6 steps. Nominal values of kV and mA were
corrected using measured values acquired using an Unfors Xi system prior to plotting the data.
Results and Discussion
The effect of varying tube voltage was investigated by varying the nominal tube voltage from 70 to 140 kVp while
leaving all other parameters unchanged. The standard CBDI is presented as a function of tube potential in figure 1.
0
5
10
15
20
25
70 90 110 130
nominal kVp
do
se -
CB
DI
(mG
y) Std weighted
Bakalyar
Figure 1: CBDI as a function of nominal kVp shown for both the
standard CBDI weighting and for the Bakalyar weighting
The low contrast visibility and standard deviation of a uniform 4 cm3 region of interest were evaluated using the
Catphan phantom and are shown in figures 2 and 3 as a function of dose. These figures show that while there is a
three-fold increase in dose changing from 90 to 140kVp, there is little improvement in the image noise, as measured
by standard deviation, beyond 90 kVp. The low contrast resolution improves with increased dose but this
improvement stabilizes somewhat beyond 90 kVp.
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
1.7
0 5 10 15 20 25Dose CBDIw (mGy)
Low
contr
ast
vis
ibil
ity (
%)
0
10
20
30
40
50
60
70
80
90
100
0 5 10 15 20 25Dose CBDIw (mGy)
SD
Figure 2: Low contrast visibility as a function of dose
variations due to changes in kVp.
Figure 3: Standard deviation as a function of dose
variations due to changes in kVp.
When tube potential is kept constant and only the tube current (mA) is varied, we find, as expected, a linear
relationship between dose and tube current. The low contrast visibility shows that resolution improves as the tube
current is increased. Figure 4 shows low contrast visibility vs dose and it is also seen that improvements in image
resolution are limited with increases in tube current beyond a dose of about 20 mGy. As with varying the voltage,
the image noise decreases sharply but shows little improvement beyond the initial drop. This is seen in figure 5
which plots standard deviation vs dose for three nominal tube potentials.
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0 10 20 30 40 50 60
Dose CBDIw (mGy)
Lo
w c
on
tras
t v
isib
ilit
y 140kV
120kV
100kV
10
12
14
16
18
20
22
24
26
28
30
0 10 20 30 40 50 60
Dose CBDIw (mGy)
Sta
ndar
d d
evia
tion
100kV
120kV
140kV
Figure 4: Low contrast visibility as a function of dose
variation due to changes in mA. Shown for three
fixed settings of kVp.
Figure 5: Standard deviation versus dose variation
due to changes in mA. Shown for three fixed settings
of kVp.
In XVI v4.5, the software allows variation in the gantry speed which in turns controls the number of projections that
will be used in the reconstruction. Keeping all other parameters constant, the dose increases linearly with the number
of projections. Figure 6 shows the low contrast resolution as a function of the number of projections. As the number
of projections increases, the low contrast visibility improves in spite of an increase in image noise as shown in
Figure 7.
1.0
1.1
1.2
1.3
1.4
1.5
1.6
0 200 400 600 800 1000 1200 1400
Number of projections
Low
contr
ast
vis
ibil
ity
12
13
14
15
16
17
18
300 500 700 900 1100 1300
Number of projections
Sta
nd
ard
dev
iati
on
Figure 6: Low contrast visibility as a function of
number of projections
Figure 7: Standard deviation as a function of number
of projections.
In each of the above cases, there was little or no change observed in the high contrast resolution as each of the
parameters was changed.
Conclusions
The dose from cone beam CT has been previously studied and these measurements confirm expectations that the
dose will increase linearly with increasing tube current and number of projections. The increase in dose as a
function of tube potential follows an exponential curve, indicating that the trade off for increasing tube voltage is
higher beyond 90 or 120kVp. Improvements in image quality are generally achieved by increasing the dose, but the
rate of improvement diminishes as the dose gets to the higher end of the range investigated. This suggests that in
order to improve image quality, increases in kV or mA beyond standard clinical settings may not have a large
enough effect on image quality to justify the additional patient dose. As treatment techniques rely more heavily on
image guidance, care must be taken to establish clinical imaging protocols. Given the clear relationship between
tube settings and dose, it must be stressed that changes to these parameters should be limited to cases where the
image quality is insufficient at standard settings and where changes to the tube settings result in a clear improvement
image quality. If no gain is achieved in the image resolution by increasing the dose, a lower setting should be used.
Current status of the NRC primary standard for 192
Ir HDR brachytherapy sources
E Mainegra-Hing*, Brad Downton,
National Research Council of Canada
Introduction
The NRC primary standard for 192
Ir HDR brachytherapy sources is revised. The NRC has offered calibration
services for these brachytherapy sources since 2011. The initial standard made use of the 7-distance technique and
the cone-shadow method to determine the source-to-detector distance offset and the room scatter respectively. After
removing the room scatter at each distance, the position offset was estimated by simple inspection until a small
enough variation in the source strength Sk was achieved. This approach suggested the room scatter to vary with
distance and no uncertainty estimate was possible for the position offset determination. As a consequence, a rather
conservative estimate of the total uncertainty was used. The present work relies on the multiple-distance method and
a non-linear weighted least-squares-fit to determine all the unknown quantities, including Sk, under the assumption
of constant room scatter. This assumption has been shown in the literature to apply when the measuring setup is
placed far enough from walls, ceiling and floor. The least-squares-fit provides the uncertainty of the estimated
parameters allowing for more realistic uncertainty budget estimation.
Methods
The 192
Ir HDR microSelectron V2 seed at NRC is mounted on an HDR afterloader on loan from Nucletron. Air-
kerma rate for this source is measured using a graphite-walled spherical ionization chamber (2S). The inverse of IrNk
is obtained as the arithmetic mean of the inverse of the Nk values for a 137
Cs beam and a medium filtered 250 kV x-
ray beam quality (N250). Measurement of the source output at several source-detector distances can be used to
determine the source strength, room scatter and positioning offset by means of a non-linear least-squares-fit, under
the assumption of constant room scatter.
The measured signal, Mraw, corrected for attenuation and scatter along the beam path and for point source divergence
(Kondo-Randolph), can thus be fitted to the expression
.
Where c is the positioning offset, Mroom, the room scatter, and f is the source strength Sk per unit Nk. The use of this
analytical expression for Mraw allows determining the covariance matrix and hence estimating the statistical
uncertainties in the parameter estimation.
Results and Discussion
EGSnrc Monte Carlo (MC) simulations of the room scatter for a VariSource spectrum in a room with same
dimensions as the NRC brachytherapy room confirm the assumption of room scatter constancy as shown on the left
panel in Figure 1. The right panel of Figure 1 shows a similar result for a bare 192
Ir source and a smaller room.
Figure 1. MC estimated room scatter in the actual room used at NRC (left) and for a smaller room (right).
Mraw must be corrected for attenuation and scatter along the beam path since the definition of source strength
requires air-kerma in vacuum. Previous studies have reported these corrections to almost cancel out. This could be
explained by the fact that most 192
Ir lines are above 60 keV, where most of the interactions in air are Compton
scattering events with small scattering angles. EGSnrc MC determination of Aatt and Ascat for a VariSource
spectrum is compared to results by Rasmussen et al (2007) using MCNP05 in Figure 2. As can be seen the product
of these quantities is almost unity.
Figure 2. Comparison of MC estimation of the attenuation and scatter corrections Aatt and Ascat using EGSnrc and
MCNP05.
Results of the non-linear least-squares fit of Mraw using Grace are shown in the left panel of Figure 3. On the right
panel of this figure, the contribution of the lead wedge to the measured signal is estimated using MC simulation. As
can be seen the use of this technique would require correcting for this contribution which is a much larger effect that
the room scatter and positioning offset.
Figure 3. Non-linear least-squares-fit results (left). Lead wedge effect (right).
If no correction is applied to the measured signal Mraw, a deviation from the inverse distance squared can be
observed in the red curve of Figure 4. Furthermore, after applying all required corrections, an almost perfect inverse
distance squared behavior is restored (black curve). Using the presented methodology in this work, the NRC
determined value for Sk is within 0.03% of the manufacturer’s reported value.
Figure 4. Uncorrected and corrected measured signals.
It is worth mentioning that the overall effect of all corrections is about 0.6%. Due to the averaging of Sk at all
distances when using the shadow-cone method, no significant difference is observed between the previous approach
the approach proposed here.
Conclusions
Under the assumption of constant room scatter, an analytical model can be used to directly estimate the required
quantities for the determination of the source strength given by the air-kerma rate in vacuum times distance squared.
The major source of uncertainty comes from the 0.45% uncertainty in the Nk values for 137
Cs and the x-ray beam
quality. The uncertainty in f from the non-linear least-squares-fit is about 0.4% and thus the statistical uncertainty in
Sk is about 0.6% (one sigma). We have chosen to only include the uncertainties which are directly related to the
NRC determination of the source strength.
Label-free Raman spectroscopy of single tumour cells detects early radiation-induced glycogen synthesis associated with increased radiation resistance
Q Matthews*1, M Isabelle2, S Harder2, AG Brolo3, JJ Lum1, and A Jirasek2
(1) BC Cancer Agency – Vancouver Island Centre (2) Physics and Astronomy, University of Victoria (3) Chemistry, University of Victoria
Introduction Altered cellular metabolism is a hallmark of tumor cells1 and contributes to a host of properties that are associated with resistance to conventional therapies, including radiation. Detection of radiation-induced biochemical changes can reveal specific metabolic pathways affecting radiosensitivity that may serve as attractive therapeutic targets. Novel technologies enabling the pre-treatment characterization and early monitoring of metabolic processes within tumour cells may provide significant opportunities for personalizing combined modality radiation therapy treatments in order to maximize the probability of disease response. One technique that has demonstrated great promise for metabolic analysis of tumour radiation response is Raman spectroscopy (RS).2-4 RS is an optical technique that allows the characterization of the species presented in complex media by the vibrational signature of each molecular component. An optical laser is focused onto a sample, and the scattered Raman photons are collected and passed through a spectrometer for analysis. An advantage of RS is that live cells or tissues can be probed without requiring any fixation, staining, or molecular targeting (label-free). RS of intact cells can provide molecular information at levels of accuracy and sensitivity comparable to other established techniques such as magnetic resonance spectroscopy and flow cytometry.5 The molecular specificity of RS allows the simultaneous detection of signals from proteins, nucleic acids, lipids, and carbohydrates (e.g., glycogen) in a single acquisition, allowing complex molecular changes in cells to be analyzed simultaneously across different classes of biomolecules and therefore bypassing certain difficulties inherent to staining or molecular targeting methods. Furthermore, RS can be directly applied in vivo with the use of minimally invasive fiber-optic probes.6 Recent work2-4 demonstrated that single-cell RS techniques applied to cells irradiated in vitro with single high doses of radiation (15 to 50 Gy) can detect radiation-induced molecular and metabolic changes in human tumour cell lines. Using principal component analysis (PCA), radiation-induced changes were distinguished from concurrent changes arising from cell cycle processes.3,4 Furthermore, RS radiation response signatures were shown to segregate the cell lines tested according to radiosensitivity and p53 gene status.4 The aims of the present study are threefold: (1) to extend these previous RS methods to clinically relevant doses (2 to 10 Gy) using both radioresistant and radiosensitive tumour cell lines, (2) to demonstrate early RS detection of radiation-induced glycogen synthesis in radioresistant cell lines, and (3) to co-treat the radioresistant cells with the anti-diabetic drug metformin to demonstrate that early RS monitoring of radiation-induced glycogen synthesis correlates with the radiosensitizing effect of the combined modality treatment. Methods Three human tumour cell lines, two radioresistant (H460, SF2 = 0.57 and MCF7, SF2 = 0.70) and one radiosensitive (LNCaP, SF2 = 0.36), were irradiated to 2, 4, 6, 8 or 10 Gy with single fractions of 6 MV photons. In additional experiments, H460 and MCF7 cells were irradiated 1 hour after incubation with 5 mM of the anti-diabetic drug metformin. Treated and control cultures were analyzed with RS daily up to 3 days post-treatment. Single-cell Raman spectra were acquired from 20 live cells per sample using previously described techniques,2 and experiments were repeated in triplicate. The combined data sets (up to 3240 cell spectra per data set) were post-processed2 and analyzed with principal component analysis using standard algorithms. Cells from each culture were also subjected to standard assays2,8 for viability, proliferation, cell cycle distribution, and radiation clonogenic survival. Results and Discussion RS detection of early radiation-induced glycogen synthesis in radioresistant tumour cells: Single-cell Raman analysis of 2-10 Gy irradiated H460, MCF7 and LNCaP cells at 1 to 3 days post-irradiation revealed radiation-induced synthesis of glycogen in H460 and MCF7 cells, but not LNCaP cells. Figure 1A shows representative irradiated and unirradiated H460 cell Raman spectra collected at 3 days post-irradiation. The point-by-point difference spectrum and the first PCA component from the entire Raman data set are both dominated by Raman spectral features of glycogen (solid black trace in Figure 1A). The first PCA component explains 40.9% of the total variance, and represents the variability in intra-cellular glycogen content within the complete Raman data set of 3240 single-cell spectra. The mean PCA scores for the first PCA component (Figure 1B) indicate that statistically significant (p<0.05 by unpaired two-tailed t-test) increases in intra-cellular glycogen, relative to same-day unirradiated cells, occur for all radiation doses at days 1-3 for H460 cells and at days 2-3 for MCF7 cells, but not at
any day for LNCaP cells. Radiation effects on proliferation, cell death, and cell cycle redistribution were similar for each cell line (data not shown). However, radiation clonogenic survival assays indicated that both H460 and MCF7 cells are significantly more radioresistant than LNCaP cells (Figure 1C). As such, we hypothesize that radiation-induced glycogen synthesis is a biomarker for increased radioresistance in human tumour cells.
Metformin co-treatment of radioresistant tumour cells: One hour prior to irradiation, H460 and MCF7 cells were incubated with 5 mM of the anti-diabetic drug metformin, which has previously been shown to radiosensitize MCF7 tumour cells via activation of signaling pathways also known to inhibit glycogen synthesis.7 Co-treatment with metformin had little effect on radiation-induced glycogen synthesis in H460 cells (Figure 2A), whereas in MCF7 cells glycogen synthesis was dramatically reduced (Figure 2B). A representative reduction in the RS glycogen signal for co-treated MCF7 cells is shown in Figure 2C. Co-treatment in H460 cells had no deleterious effects on cell proliferation or viability, whereas co-treated MCF7 cells exhibited significantly reduced proliferation and increased cell death (Figures 2D and 2E). Finally, clonogenic assays demonstrated no effect of metformin co-treatment on the radiosensitivity of H460 cells (Figure 2F), whereas MCF7 cells were significantly radiosensitized (Figure 2G). Conclusions Label-free RS is well suited for early detection of glycogen synthesis post-irradiation, a previously undocumented metabolic mechanism that is (1) associated with tumour cell radioresistance, and (2) can be targeted to increase radiosensitivity. In this work, RS monitoring of radiation-induced glycogen synthesis was found to correlate with the efficacy of the targeted co-treatment strategy. RS monitoring of intratumoral glycogen levels in radiotherapy patients may provide new opportunities for personalized combined modality treatments.
Fig. 1 (A) Single-cell Raman spectra of an irradiated (10 Gy) and unirradiated H460 cell at 3 days post-irradiation. The difference spectrum (dashed trace) is shown for comparison with the first PCA component (solid gray trace) from the entire RS data set, and the Raman spectrum of glycogen (solid black trace). (B) Mean PCA scores (N=60 spectra per point) for the first PCA component. (C) Clonogenic survival of irradiated H460, MCF7 and LNCaP cells. *** - Curves significantly different with p<0.0001 by extra sum-of-squares F test. Error bars are standard error on the mean of three independent experiments.
References 1 D. Hanahan and R. A. Weinberg (2011). Cell 144:646-674. 2 Q. Matthews et al. (2010). Applied Spectroscopy 64:871-887. 3 Q. Matthews et al. (2011). Physics in Medicine and Biology 56:19-38. 4 Q. Matthews et al. (2011). Physics in Medicine and Biology 56:6839-6855. 5 J. R. Mourant et al. (2006). Journal of Biomedical Optics 11:064024. 6 J. C. C. Day et al. (2009). Physics in Medicine and Biology 54:7077-87. 7 C. W. Song et al. (2012). Scientific Reports 2:362. 8 N. A. P. Franken et al. (2006). Nature Protocols 1(5):2315-2319.
Fig. 2 (A&B) Mean PCA scores (N=60 spectra per point) for the first PCA component from metformin co-treatment on (A) H460 and (B) MCF7 cells. (C) Single-cell Raman spectra of 10 Gy irradiated MCF7 cells at 3 days post-irradiation, with and without 5 mM metformin. The difference spectrum (dashed trace) is shown for comparison with the first PCA component (solid gray trace) from the entire RS data set, and the Raman spectrum of glycogen (solid black trace). (D) Number of viable cells relative to controls, and (E) percentage dead cells in each culture, at 3 days post-treatment. (F&G) Clonogenic survival of metformin co-treated (F) H460 and (G) MCF7 cells. n.s. – no significant difference between curves, *** - curves significantly different with p<0.0001 by extra sum-of-squares F test. Error bars are standard error on the mean from three independent experiments.
Unified Optimization and Delivery of Intensity-modulated Radiation Therapy and Volume-modulated Arc
Therapy
J Chen*1,2,4
, M MacFarlane4, E Wong
1,2,3, D Hoover
1,2,4
(1) Department of Oncology, (2) Department of Medical Biophysics, (3)Department of Physics and Astronomy,
University of Western Ontario,(4) London Health Science Centre, London, ON, Canada
Introduction
Volumetric-modulated arc therapy (VMAT) has been rapidly adopted by the radiotherapy community due primarily
to its delivery speed and monitor unit (MU) efficiency, as well as the conformal dose distributions it can achieve [1].
On the other hand, intensity-modulated radiotherapy (IMRT) with its static beam directions might be advantageous
in cases where steep dose gradients or highly intensity-modulated beam intensities are required in certain preferred
directions [2]. While the community tends to regard these two delivery techniques as disparate entities, they are in
reality special cases of one another. More specifically, there exists a unifying delivery technique which bridges the
gap between VMAT and static-gantry IMRT. Such a unified delivery, if properly implemented into an inverse-
planning algorithm, would in general lead to improved dose delivery capabilities as the algorithm could naturally
tune the beam within a given arc range to be more IMRT-like, if increased modulation is required, or more VMAT-
like, if increased conformity is required. The purpose of this work is to study the feasibility of a unified intensity-
modulated arc therapy (UIMAT) that combines IMRT and VMAT optimization and delivery in the same arc for
producing efficient and superior radiation treatment plans.
Methods
The optimization of UIMAT was started by creating static beams uniformly spaced at certain degree increments (24
degrees was used for this initial study) between the user-selected start and stop angles. IMRT objectives were then
created using the standard Pinnacle inverse-planning user interface (Philips Medical System), after which fluence
optimization was initiated. Conversion to deliverable MLC segments was then carried out with appropriate
conversion parameters. After creating MLC segments, a direct machine parameter optimization (DMPO) step was
performed. After this step, customized software was used to redistribute these control points into a UIMAT beam.
During UIMAT conversion, any beam with four or more control points was converted into a slow-moving arc with
0.1 degree control point spacing (termed the IMRT phase). The remaining beams were joined into multiple partial
arcs with standard 4 degree control point spacing, representing the standard VMAT phase. Certain “soft”
deliverability constraints, for example the maximum MU per degree, were relaxed for the IMRT-like portions of
delivery. This was required in order to have a reasonable number of MUs delivered during the IMRT phase, which
may have up to five control points within a single degree spacing. It is important to note that such a beam is still
machine-deliverable as it does not violate any physical constraints. From this point on, optimization proceeds using
the standard functionality within Pinnacle. DMPO optimization is continued until an optimized plan is obtained.
Five treatment plans each for prostate, head and neck, and lung were generated using our UIMAT technique and
compared with clinical VMAT or IMRT plans. Delivery verification was performed on an ArcCheck phantom and
delivered in clinical mode on a Varian TrueBeam linear accelerator.
Results
The UIMAT plans were generated for 15 cases as shown in Table 1. In general, UIMAT uses only one arc instead of
two arcs compared with VMAT plans. The numbers of MLC control points are less than VMAT plans but more than
IMRT plans. The estimated delivery times for UIMAT plans are similar to VMAT plans and are expect to be faster
than multiple-field IMRT plans due to extra time required to mode up individual IMRT field. As a example,
comparison of dose distributions and DVHs between VMAT and UIMAT plans for a head and neck case are shown
in Figure 1 and Figure 2 respectively. The dosimetric parameters are given in Table 2, showing UIMAT plans have
lower doses for most OARs compared to VMAT plans with similar dose coverage to PTVs.
Table 1: General and beam information about the patients in each group.
No. Site Dose [Gy]
Clinical Beam Characteristics
UIMAT Beam
Number of Control Points
Estimated Delivery Time [s]
Clin UIMAT Clin UIMAT
1 Lt Parotid 64/60/54 2x 210o VMAT 1x 210
o 108 65 91 64
2 Rt Parotid 60 2x 225o VMAT 1x 225
o 116 55 95 68
3 Larynx/Neck 70/56 2x 360o VMAT 1x 360
o 182 87 151 171
4 Neck /Parotids 70/56 2x 360o VMAT 1x 360
o 182 89 151 201
5 Larynx 61/50 5 Field IMRT 1x 260o 23 77 ---- 109
6 Lt Lung 60 5 Field IMRT 1x 230o 17 73 ---- 64
7 Lt Lung 60 2x 225o VMAT 1x 225
o 116 63 94 67
8 Rt Lung 60 6 Field IMRT 1x 192o 21 61 ---- 126
9 Rt Lung 60 2x 210o VMAT 1x 210
o 108 63 90 98
10 Lt Lung/Med. 50 2x 360o VMAT 1x 360
o 181 93 149 100
11 Prostate 76 1x 360o VMAT 1x 360
o 91 99 79 129
12 Prostate Bed 66 2x 360o VMAT 1x 360
o 182 97 151 200
13 Prostate 45 2x 360o VMAT 1x 360
o 182 103 150 236
14 Prostate 76/50.4 2x 360o VMAT 1x 360
o 182 93 154 241
15 Prostate Bed 66 2x 360o VMAT 1x 360
o 182 96 151 139
Abbreviations: Clin = Clinical; Rt = Right; Lt = Left
Figure 1. Dose distribution comparison between VMAT (left) and UIMAT (right) plans for a head and neck case.
Figure 2. DVH comparison between VMAT and UIMAT plans for a head and neck case.
Table 2. Dose volume parameters for five head-and-neck cases, comparing the clinical VMAT or IMRT plans with UIMAT
plans.
No. Lt Parotid
[Gy]
Rt Parotid
[Gy]
Oral Cavity
[Gy]
Cord D0.1cc
[Gy]
CI [Gy]
Body V105%
[cm3]
Clin. UIMAT Clin. UIMAT Clin. UIMAT Clin. UIMAT Clin. UIMAT Clin. UIMAT Clin. UIMAT
11 6.9 2.4
-- --
28.4 22.6 21.5 19.5
0.21 0.77 0.63
0.14 0.79 0.63
63.9 60.9 55.9
65.2 61.6 56.0
0 0
12 5.9 3.3 61.0 60.8 30.0 27.7 35.0 34.1 0.85 0.80 60.8 60.9 1.5 2.6
13 25.9 23.1 25.6 22.4 33.2 30.0 36.3 41.0 0.83 0.79
0.70 0.70
70.0 57.0
71.2 57.5
0 0.2
14 25.3 22.5 25.4 22.9 39.3 37.3 38.5 43.9 0.78
0.72
0.82
0.71
70.4
58.5
70.5
58.0 0 0
15 0.2 0.2 0.2 0.2 0.1 0.2 0.5 0.4 0.89
0.73
0.93
0.73
61.6
54.8
62.0
54.3 0 0
Abbreviations: : D0.1cc = minimum dose to the hottest 0.1 cm3 non-contiguous volume; V105% = volume receiving at
least 105% of the prescription dose; CI = conformity index; = mean dose; Clin = clinical plan; UIMAT = unified
intensity-modulated arc therapy plan; Rt = Right; Lt = Left.
Conclusions
In this proof-of-concept work, we demonstrated that a novel radiation therapy delivery technique UIMAT which
combines VMAT and IMRT delivery in the same arc is feasible. Initial results showed UIMAT has the potential to
be superior to either standard IMRT or VMAT.
References
[1]. K. Otto, Volumetric modulated arc therapy: IMRT in a single gantry arc. Med Phys, 35(1):310–317, Jan 2008.
[2]. X. Jiang, T. Li, Y. Liu, L. Zhou, Y. Xu, X. Zhou and Y. Gong, Planning analysis for locally advanced lung
cancer: dosimetric and efficiency comparisons between intensity-modulated radiotherapy (IMRT), single-arc/partial-
arc volumetric modulated arc therapy (SA/PA-VMAT), Radiation Oncology, 6:140, 2011.
Solid line = VMAT
Dashed line = UIMAT
Optimizing planning target volume in lung radiotherapy using deformable registration
P Hoang*1, M Wierzbicki
1,2
(1) *McMaster University, Medical Physics & Applied Radiation Sciences Department, Hamilton, Ontario
(2) Juravinski Cancer Centre, Medical Physics Department, Hamilton, Ontario
Introduction
The overall goal of radiation therapy (RT) is to maximize the dose to the tumour while minimizing dose to the
surrounding normal tissue. Highly conformal radiation dose distributions can be achieved with techniques such as
intensity modulated radiation therapy (IMRT). However, sources of treatment uncertainties include patient set-up
variability, inter and intra-fractional organ motion [1], and organ deformation [2]. Intrafractional motion due to
breathing is addressed using four-dimensional computed tomography (4DCT) [3]. In one approach, the 4DCT is
used to define the internal gross tumour volume (IGTV), the region that encompasses the entire GTV throughout the
respiratory cycle. The IGTV is then expanded for subclinical disease to obtain the internal target volume (ITV).
Other uncertainties including patient set-up error, anatomical motion that is not due to breathing and machine
inaccuracies are addressed by adding a margin around the ITV to obtain the planning target volume (PTV).
The use of image guided radiotherapy (IGRT) has the potential to reduce margins, in turn reducing radiation
toxicity and allowing dose escalation which has shown to improve local control and survival [4]. For example, the
advent of linear accelerator-mounted cone-beam computed tomography (CBCT) has provided volumetric images to
address tumour position and surrounding organs at risk [5]. Following patient set-up, CBCT images are registered
with the planning CT using automated rigid image registration, and the resulting x, y, z shift is applied to the
treatment couch to correct patient position prior to treatment. Despite these advancements, smaller PTV margins
come at the expense of increasing the chance of a geometrical miss [1]. It is important to optimize PTV margins to
ensure sufficient target coverage with minimal exposure of normal tissue.
Methods
This study analyzed data from 18 patients who received conventional lung RT with CBCT image guidance at our
institution from January 2012 to September 2013. When tumour motion due to breathing was a concern, a 4DCT
was acquired along with a free-breathing planning image using a Phillips Brilliance Big Bore CT scanner (Phillips
Healthcare, Andover, MA) with 3.0 mm slice thickness. A maximum intensity projection (MIP) image was also
generated from the reconstructed 4DCT dataset.
The ITV was contoured by the radiation oncologist using the MIP image. To safely account for patient set-up
error and daily anatomical changes, a 1 cm margin was added to the ITV to obtain the PTV. Three-dimensional
conformal or intensity modulated radiation therapy plans were developed using Pinnacle v9.2 (Phillips Healthcare,
Andover, MA) and were optimized such that 95% of the PTV receives at least 90% of the prescription dose.
All patients were treated on Varian Clinac (Varian Medical Systems, Inc., Palo Alto, CA) linear accelerators
equipped with the on-board imaging (OBI) system for kV CBCT acquisition [low dose thorax mode was used].
Daily pre-treatment CBCT images were obtained after patient set-up. Rigid image registration of the spinal anatomy
between the pre-treatment CBCT and planning CT was performed. Each result was manually assessed to ensure
matching was accurate along the spinal cord and carina, and that the visible target was within the PTV. The patient
was repositioned if the translational and rotational differences between planning and treatment exceeded 15 mm and
5 degrees, respectively. Otherwise, translations between 2 and 15 mm were recorded and a translational couch shift
was applied prior to treatment. A weekly post-treatment CBCT image was also acquired.
Polygonal meshes were generated from the contours of the ITV in the treatment planning system (TPS) for each
patient to represent the surface of the ITV. The clinical couch shifts were calculated as the difference between the
couch positions (longitudinal, lateral, and height) of the post and pre-treatment CBCT images. This couch shift was
applied to the previously obtained pre-treatment CBCT to obtain a set of data for each fraction now consisting of
three CBCT images: 1. pre-couch shift; 2. pre-treatment, post-couch shift; 3. post-treatment.
Deformable image registration (DIR) was used to register the planning data to each CBCT image. First, the
average planning dataset was globally registered with the CBCT image to account for rotations, translations, and
scalings. A previously validated DIR algorithm [6] was then used to fully register the data. The resulting
transformation from the DIR was applied to the ITV surface to obtain a deformed ITV surface for each fraction.
Treatment success was quantified by determining the percentage of the treatment ITV surface vertices within the
evaluation PTV margin. Furthermore, we determined a margin that provided a suitable level of ITV coverage over
an appropriate percentage of fractions. Statistical analyses were conducted to quantify the benefit of CBCT imaging,
effects of intrafractional motion, and optimal PTV margins.
Results
Typical results obtained using deformable image registration are shown in Figure 1 while Figure 2 demonstrates
how target coverage was assessed. Figure 3 shows the percentage of treatment fractions where at least 99% of the
ITV fell within the PTV (thus the ITV is believed to have received the prescribed dose).
Figure 1. Result of deformable image registration for patient 1, fraction 1. Axial slices from (A) 4D planning image
with ITV, (B) post-treatment CBCT with no ITV, and (C) deformed planning image with treatment-specific ITV.
Figure 2. Assessment of tumour coverage following the localization of the ITV onto the CBCT image of patient 1,
fraction 1. A. Original planning ITV (yellow) expanded by a 5 mm isotropic margin to obtain the PTV (white). B.
Adapted ITV following initial setup showing deformation and a geometric miss. C. Adapted ITV following couch
correction showing reduced geometric miss. D. Adapted ITV following treatment completion showing similar
coverage to C. This example illustrates a case where image-guidance was successful in improving geometrical
coverage following image registration; however, a percentage of the ITV is still shown to be outside of the PTV in
both pre- and post-treatment cases for this particular slice using a 5-mm PTV margin.
Figure 3. Percentage of fractions with sufficient coverage
versus isotropic PTV margin, where sufficient coverage was
defined by at least 99% ITV coverage. Analysis was
conducted on 79 fractions across 18 patients. Red bars
represent the coverage without image guidance and blue bars
represent the situation with CBCT image guidance.
Discussion
Our treatment success was defined as a situation where at least 99% of the ITV is covered by the PTV. Although this
criterion may be arbitrarily chosen, it is supported by a study conducted by Guckerberge et al. [7] who proposed a
PTV margin such that tumour drifts were less than a certain tolerance in 90% of all fractions. Actually, this criterion
appears to be conservative as indicated by the results of van Sornsten de Koste et al. [8] which showed sufficient
dosimetric coverage for geometrical coverage less than 99% accuracy.
Figure 3 demonstrates the importance of CBCT image guidance. The current approach with a 1 cm PTV margin
was successful ~96% of the time. A 0.8 cm isotropic margin was successful ~92% of the time. A 0.5 cm margin
only contained the ITV in ~ 75% of the fractions in contrast to the zero geometrical misses using a 0.5 cm isotropic
PTV margin observed by Higgins et al. [9] This difference may be attributed to the capability of DIR to detect
smaller geometrical changes (Figures 1 and 3) compared to the rigid registration analysis performed by Higgins et
al. [9]
Non-isotropic margins were also evaluated. The 0.6 x 0.6 x 1.0 cm3 margin (0.6 cm in plane, 1 cm craniocaudal)
was found to be successful ~91% of the time. This margin demonstrated statistically significant accuracy
improvement when CBCT imaging was employed and demonstrated insignificant losses in accuracy during the
treatment fraction.
This study presents a comprehensive method for retrospectively analyzing IGRT data to optimize treatment
margin. Applying this analysis to our clinical data shows that a 0.8 cm isotropic or 0.6 x 0.6 x 1.0 cm3
non-isotropic
PTV margin is appropriate. The advantage of the latter is that it allows treatment of targets closer to the esophagus
and spinal cord.
References
[1] van Herk M. Errors and margins in radiotherapy. Semin Radiat Oncol. 2004;14(1):52-64.
[2] Yan D, Jaffray D, Wong J. A model to accumulate fractionated dose in a deforming organ. Int J Radiat Oncol
Biol Phys. 1999;44(3):665-675.
[3] Rietzel E, Pan T, Chen GT. Four-dimensional computed tomography: Image formation and clinical protocol.
Med Phys. 2005;32(4):874-889.
[4] Boda-Heggemann J, Lohr F, Wenz F, Flentje M, Guckenberger M. kV cone-beam CT-based IGRT. Strahlenther
Onkol. 2011;187(5):284-291.
[5] Bissonnette J, Purdie TG, Higgins JA, Li W, Bezjak A. Cone-beam computed tomographic image guidance for
lung cancer radiation therapy. Int J Radiat Oncol Biol Phys. 2009;73(3):927-934.
[6] Wierzbicki M, Drangova M, Guiraudon G, Peters T. Validation of dynamic heart models obtained using non-
linear registration for virtual reality training, planning, and guidance of minimally invasive cardiac surgeries. Med
Image Anal. 2004;8(3):387-401.
[7] Guckenberger M, Meyer J, Wilbert J, et al. Intra-fractional uncertainties in cone-beam CT based image-guided
radiotherapy (IGRT) of pulmonary tumors. Radiother Oncol. 2007;83(1):57-64.
[8] van Sörnsen de Koste JR, Lagerwaard FJ, Schuchhard-Schipper RH, et al. Dosimetric consequences of tumor
mobility in radiotherapy of stage I non-small cell lung cancer–an analysis of data generated using ‘slow’CT scans.
Radiother Oncol. 2001;61(1):93-99.
[9] Higgins J, Bezjak A, Franks K, et al. Comparison of spine, carina, and tumor as registration landmarks for
volumetric image-guided lung radiotherapy. Int J Radiat Oncol Biol Phys. 2009;73(5):1404-1413.