- Email: [email protected]
Evaluation of set-up errors in head and neck radiotherapy using electronic portal imaging
Physica Medica 29 (2013) 531e536
Contents lists available at SciVerse ScienceDirect
Physica Medica journal homepage: http://www.physicamedica.com
Original paper
Evaluation of set-up errors in head and neck radiotherapy using electronic portal imaging Bojan Strbac a, *, Vesna Spasic Jokic b,1 a b
International Medical Centres, Centre for Radiotherapy, Dvanaest beba bb, 78000 Banja Luka, Bosnia and Herzegovina University of Novi Sad, Faculty of Technical Science, Trg Dositeja Obradovica 6, 21000 Novi Sad, Serbia
a r t i c l e i n f o
a b s t r a c t
Article history: Received 20 April 2012 Received in revised form 28 November 2012 Accepted 7 December 2012 Available online 2 January 2013
Introduction: The aim of this study was to evaluate three-dimensional (3D) set-up errors and propose optimum margins for planning target volume (PTV) coverage in head and neck radiotherapy. Methods: Thirty-five patients were included in the study. The total number of portal images studied was 632. Population systematic (S) and random (s) errors for the patients with head and neck cancer were evaluated based on the portal images in the caudocranial longitudinal (CC) and left-right lateral (LR) direction measured in the anterior-posterior (AP) field, as well as from the images in the caudocranial longitudinal (CC) and dorsoventral lateral (DV) direction measured in the lateral (LAT) field. The values for the clinical-to-planning target volume (CTV-PTV) margins were calculated using ICRU Report 62 recommendations, along with Stroom’s and van Herk’s formulae. Results: The standard deviations of systematic set-up errors (S) ranged from 1.51 to 1.93 mm while the standard deviations of random set-up (s) errors fell in between 1.77 and 1.86 mm. The mean 3D vector length of displacement was 2.66 mm. PTV margins calculated according to ICRU, Stroom’s and van Herk’s models were comprised between 1.95 and 6.16 mm in the three acquisition directions. Discussion and conclusions: Based on our results we can conclude that a 6-mm extension of CTV to PTV margin, as the lower limit, is enough to ensure that 90% of the patients treated for head and neck cancer will receive a minimum cumulative CTV dose greater than or equal to 95% of the prescribed dose. Ó 2012 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.
Keywords: Set-up errors Radiotherapy 3DCRT IMRT Head and neck Portal images PTV margins
Introduction Cancer is the second leading cause of death in the industrialized countries and the only major disease for which death rates are increasing. The demand for cancer care is expected to increase over the coming decade. Radiotherapy plays a key role in cancer treatment. According to the statistical data from the Republic of Serbia and the Republic of Srpska (a part of Bosnia and Herzegovina), about 0.5% of the population will develop a new cancer per year and at least a half of them will require radiotherapy at some stage of their illness. Among the patients having radiotherapy, more than 65% are treated with curative intent, often in combination with surgery and chemotherapy. In addition, radiotherapy has an important role in the palliation of cancer symptoms.
* Corresponding author. Tel. þ387 65859068. E-mail addresses: [email protected], [email protected] (B. Strbac), [email protected] (V.S. Jokic). 1 Tel.: þ381 21 485 2569.
Radiotherapy treatment is a local one so the main goal of radiation therapy planning is to maximize the dose to the target while limiting the dose to nearby healthy organs (“risk organs”), in order to improve the control of tumor growth and to reduce side effects. Evaluation of treatment-related errors, especially those resulting from patient set-up and organ motion, is an increasingly important part of the clinical radiotherapy process. The linear accelerator with multileaf collimation technology has evolved to such an extent that radiation dose can be delivered with high accuracy to volumes in regular objects. Unfortunately, the accuracy of dose delivery to dynamic cancer targets is limited by the uncertainty in many of the treatment parameters. A treatment uncertainty usually includes: an uncertainty in organ shape and motion, beam geometry and patient set-up. Inter-fractional (day-to-day) geometric change occurs over the weeks of therapy, due to digestive processes, change in breathing patterns, differences in patient set-up, and treatment responses such as the growth or the shrinkage of the tumor or the nearby risk organs (e.g., the parotids in head and neck treatment). These uncertainties are taken into account by population-based “uncertainty” margins around the target area, which may be excessive or conservative and are applied to the structures identified
1120-1797/$ e see front matter Ó 2012 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejmp.2012.12.001
532
B. Strbac, V.S. Jokic / Physica Medica 29 (2013) 531e536
before initiation of the therapy [4]. A reduction in treatment set-up errors can be achieved by improving laser alignment, table and gantry stability and accuracy, and the patient set-up procedures by the use of appropriate immobilization methods [3]. Each set-up error consists of a systematic component, i.e. the same deviation in the same direction for each fraction throughout the whole course of treatment, as well as a random component, varying from day-to-day [4]. Both types of errors were investigated for 35 head and neck radiotherapy (HN-RT) patients by applying an electronic portal imaging device (EPID). All patients included in this report underwent 3D conformal and intensity modulated head and neck radiotherapy (3DCRT, IMRT) in our institute. There are three main types of volume defined in radiotherapy planning. The first one is given by the position and extent of gross tumor, i.e. what can be seen, palpated or imaged, and is defined as the gross tumor volume (GTV). The second volume contains the GTV, plus a margin reflecting sub-clinical disease spread which therefore cannot be fully imaged; this is known as the clinical target volume (CTV) [1]. The CTV is important because this volume must be appropriately treated to achieve cure. The third volume, the planning target volume (PTV), allows for uncertainties in planning or treatment delivery. It is a geometric concept designed to ensure that the radiotherapy dose is actually delivered to the CTV [1]. Radiotherapy planning must always consider critical normal tissue structures, known as organs at risk (ORs). In some specific circumstances, it is necessary to add a margin analogous to the PTV margin around an OR in order to ensure that the organ cannot receive a higher-than-safe dose; this gives a planning organ at risk volume [24]. Accurate positioning in the radiation treatment is important because set-up errors can result in significant underdose to the tumor and/or overdose to one or more critical organs [2]. The first aim of this study was to estimate systematic and random set-up uncertainties by means of portal imaging for patients with head and neck cancer treated with 3DCRT and IMRT technique, and to find optimal CTV to PTV margin expansion. The second aim of the study was to estimate the impact of the frequency of online verification on set-up accuracy and set-up margins. Materials and methods Thirty-five patients were included in this study. Twenty-nine of them underwent 3D conformal radiotherapy of head and neck region, with at least one weekly image (patients ¼ 29) whereas the remaining six underwent IMRT with daily imaging (patients ¼ 6). HN-RT patients were immobilized using 5-point Orfit masks (four fixation points plus an additional one on the top of the head). After a CT scan of the patient was obtained, DRRs of the beam’s eye view (BEV) were computed and stored in the treatment planning
system (TPS) and were successively considered to be the reference images. The DRR images were saved in a database accessible to our software. In our clinic we use ARIA (Varian Medical System) as the information and image management system that aggregates patient data into a fully-electronic medical chart. Portal images were acquired by using the EPID system attached to Varian linear accelerators. To ensure accuracy on the treatment couch, patient positioning adjustments were made aligning surface markers (i.e. tattoos) placed at the time of simulation. Radiation therapists (four of them) visually compared the portal images to planned DRRs by manually matching bone anatomy landmarks by means of anatomy matching software PortalVisionÔ Advanced Imaging (Varian Medical System). The therapists had a full control over image display contrast, window level, image filter and magnification. For patients who were treated by 3DCRT, the measurement of set-up errors was performed at least three times during the first week of treatment and on a weekly basis thereafter. Portal imaging was done before the treatment and set-up errors were assessed. The offline correction protocol was created so as to correct set-up errors should the mismatch have been found to be more than 3 mm in any direction. No action was taken if the mismatch was within 3 mm in all direction. If the PIs were incorrect, the patient position was corrected within the mask and the additional PI was performed to recheck the positioning [6]. The corrective shifts (in table parameters) were applied and kept constant every day for subsequent treatment until the next imaging was done, typically the following week. If there were more than one portal imaging in a day, the setup errors were analyzed from the first and last portal of the day. Compared to 3DCRT radiotherapy, sharp dose gradients are believed to make Intensity Modulated Radiotherapy (IMRT) dose distribution more sensitive to geometric uncertainties [5]. This is a major concern, as the delivered dose to the patient may deviate significantly from the approved plan, not only preventing the benefits of IMRT over 3DCRT from being clinically realized, but also potentially resulting in the delivery of a worse treatment. For all these reasons, all the head and neck IMRT patients at our clinic were treated using an image-guided radiation therapy technique based on daily orthogonal portal images. Set-up deviations were measured in 318 anterior and 314 lateral portal images and were assessed in caudocranial longitudinal, and in left-right lateral direction measured in AP field as well as in dorsoventral vertical, and caudocranial longitudinal direction measured in the lateral field. Displacements, i.e. total set-up error m, were assessed offline with Varian Offline Review anatomy matching software (Fig. 1). Mean displacements, population systematic and random errors, and 3D vector of displacement were calculated.
Figure 1. Varian offline review (a) AP images; (b) Lateral images (portal and DRR).
B. Strbac, V.S. Jokic / Physica Medica 29 (2013) 531e536
There is a time gap between PIs acquired for 3DCRT and those acquired for IMRT treatment, because our institution has only recently started with IMRT treatment of patients. This is the main reason why only six patients included in this study underwent IMRT treatment. Patient set-up analysis The mismatch result, i.e. set-up deviation m, is a deviation combining both random and systematic component. For each patient, systematic and random deviations were calculated for all the three directions separately. A patient set-up deviation is defined as the difference between the actual and intended positions of the part of the patient’s body irradiated. The standard deviation of the random errors is denoted as s, defined as the standard deviation of the day-to-day set-up positions averaged over all the patients in the group. The distribution of systematic P deviations is denoted as , defined as the standard deviation of the distribution of average set-up deviations per patient [7]. The reliability of the following statistical approach which estimates these standard deviations depends on the total number of patients P and images N used in the study. All the patients are assumed to be consistent in terms of set-up technique. The calculation method assumes that both random and systematic components are normally distributed [4]. The mean deviation of np images (measurements) i.e. the individual systematic set-up deviation for a patient p is given by:
mp ¼
533
1 X m : np i ¼ 1;n ðPIDRRÞ;i
(1)
p
The individual random set-up deviation, i.e. the standard deviation of the distribution of m deviations around mp for a patient p in a given direction (inter-fractional):
srand;p
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 2 u1 X mðPIDRRÞ;i mp : ¼ t np i ¼ 1;n
(2)
p
The overall mean systematic error for all the P patients in the study:
moverall ¼
1 X np $mp ; N p ¼ 1;P
(3)
where N is the total number of images in the study. The Equation (3) is weighed according to the number of the portal images taken per patient. The random set-up error for all the P patients in a given direction i.e. the standard deviation of the srand,p distribution:
ssetup
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u X u P s2 ¼ t np 1 : N P p ¼ 1;P rand;p
(4)
Figure 2. Distribution of total deviations m, measured in the (a) caudocranial longitudinal and (b) left-right lateral direction from the anterior-posterior field (AP) as well as in the (c) caudocranial longitudinal direction and (d) dorsoventral vertical direction from lateral field (LAT).
534
B. Strbac, V.S. Jokic / Physica Medica 29 (2013) 531e536
Figure 3. Scatter plots illustrate 2D displacement vectors by crosses in the AP field and the lateral field respectively. The displacements are expressed in mm for all the patients in the study.
The standard deviation of the random treatment set-up error for all theP patients in a given direction i.e. the standard deviation for the mp distribution is as follows:
Ssetup
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u X 2 u 1 ¼ t np mp moverall : NðP 1Þ p ¼ 1;P
(5)
pffiffiffi Values of moverall greater than t$Ssetup = P would indicate a statistically significant overall systematic error at the 95% confidence limit and should be investigated (the mean deviations mp were assumed to follow a t-distribution; for a 95% confidence level and P 1 ¼ 34 degrees of freedom the tabulated t value is z 2) [4]. To calculate a 3D vector length, the measured 1D deviations in the three directions are combined quadratically. The mean 3D set-up deviations (one value per patient) are calculated as the average length of the 3D vector deviations over all measured fractions [3]. Due to a limited number of observed fractions per patient the valueSsetup will tend to be overestimated due to the presence of the random errors. De Boer et al. [8] provide a correction that can be applied to give an unbiased estimator of variance:
ðSunbiased Þ2 ¼ ðSbias Þ2
2 1 X sinter;p P P ¼ 1;P np
(6)
The above method assumes the normal distribution of random and systematic components [4]. Several mathematic formulae have been recommended for generating CTV-PTV margins. The ICRU 62 [9] states that systematic and random uncertainties should be added in quadrature, which should then be used for margin calculation. However, this approach assumes that random and systematic errors have an equal contribution to the dose distribution, which might not be necessarily true. Random errors blur the dose distribution whereas systematic errors cause a shift in the cumulative dose distribution [10]. Using coverage probability matrices and dose population histograms, Heijmen and Stroom [11] and van Herk et al. [12] have suggested formulae incorporating these differential effects. Stroom’s margin recipe ð2S þ 0:7sÞ ensures that, on average, 99% of the CTV receives more than or equal to 95% of the prescribed dose. The formula by van Herk ð2:5S þ 0:7sÞ ensures that 90% of patients in the population receive a minimum cumulative CTV dose of at least 95% of the prescribed dose [12]. To calculate a 3D vector length, the measured 1D deviations in the three directions are combined quadratically [3].
Results The overall mean systematic deviation moverall was calculated according to Equation (3). The random set-up error,ssetup i.e. the standard deviation of the srand;p for all the patients, was calculated according to Equation (4), and the systematic error Ssetup i.e. the standard deviation of mp , was calculated according to Equation (5) Fig. 2. Figure 3 shows the total deviations m, measured in the (a) caudocranial longitudinal and (b) left-right lateral direction from the anterior-posterior field (AP) along with those measured in the (c) caudocranial longitudinal direction and (d) dorsoventral vertical direction from the lateral field (LAT). The distribution appears to be a Gaussian one. Figure 3 is a graphic representation of the displacement vectors in the AP and LAT field, respectively. The results summarized in Table 1a take into account all the data acquired from daily and weekly PIs. In Table 1b an analysis of the distributions of daily and weekly acquired data is shown. The differences in standard deviations (SDs) of the overall displacement between daily and weekly portals were tested using Levene’s test for homogeneity of SDs [13] (Berrin Pehlivan, 2009). A p-value of <0.05 was considered statistically significant. The mean 3D vector length of displacement was 2.66 mm. Results for CTV-PTV margins were calculated using ICRU Report 62 as well as Stroom’s and van Herk’s formulae [9,11,12] (Table 2). Discussion The purpose of this study was to evaluate the patient set-up errors for patients with head and neck cancer. The sensitivity of a treatment plan to set-up errors is dependent on the sharpness of
Table 1a The summary of results for the population systematic S and random s errors in the patients with head and neck cancer evaluated from portal images in the caudocranial longitudinal and left-right lateral direction measured in the AP field, and dorsoventral and caudocranial field measured in the LAT field. Field
AP
Direction
Caudocranial longitudinal
Left-right lateral
LAT Dorsoventral vertical
Caudocranial longitudinal
Min deviation [mm] Max deviation [mm] moverall [mm] sset-up [mm] Ssetup [mm]
6 10 1.55 1.86 1.51
9 6 0.01 1.83 1.93
9 7 0.15 1.77 1.42
5 8 1.09 1.81 1.52
B. Strbac, V.S. Jokic / Physica Medica 29 (2013) 531e536
535
Table 1b The summaries of results for the population systematic S and random s errors for 3DCRT patients evaluated in basis of the weekly portal images and IMRT patients evaluated in basis of the daily portal images. Field
AP
Direction
Caudocranial longitudinal
Left-right lateral
Dorsoventral vertical
Caudocranial longitudinal
2.22 2.21 0.778
2.62 2.54 0.507
1.99 2.41 0.022
2.49 1.80 0.000
Daily PI Weekly PI
Overall standard deviation [mm] Levene’s p value
Lateral
the gradients in the plan. Increasing the number of beams in an IMRT treatment plan will increase the sharpness of the gradients around OARs and consequently the sensitivity to set-up errors increase [13]. Xing et al. [14] observed that a 3-mm error in the couch location in the AP direction resulted in a 38% decrease of the minimal target dose or in a 41% increase of the minimal spinal cord dose. Therefore, it is highly important to quantify and reduce patient set-up errors. In our institution we have decided to perform daily online verification in all the patients treated with IMRT and in the patients treated with 3DCRT where critical organs at risk are located in close proximity of the PTV. For patients where the prescribed dose does not exceed tolerance doses for organs at risk, the frequency of PIs is lower (weekly). We analyzed the patient set-up accuracy using the concept of systematic and random errors. The systematic component of any error can be defined as a deviation that occurs in the same direction and is of a similar magnitude for each fraction throughout the treatment course (treatment preparation errors), and the random component as a deviation that can vary in direction and magnitude for each delivered treatment fraction (treatment execution errors). The differentiation between systematic and random errors is important for the derivation of appropriate safety margins. There are a few limitations in our study, firstly concerning possible rotational errors because portal imaging in the AP and lateral projection has not captured rotational shifts. Secondly, portal images do not give information about organ motion; therefore, we have not accounted for these errors in calculating PTV margins. Suzuki et al. [23] reported that systematic and random set-up errors for organ motion ranged from 0.2 to 0.8 mm and 0.3e0.6 mm, respectively. Thirdly, a potential bias in the interpretation of the data concerning set-up errors lies in the observer’s judgment during image registration. Van Lin et al. [3] reported that the difference between two observers averaged over all patients is 0.1 mm or less for all directions. Since the absolute values of overall mean systematic error for patient set-up deviation in the longitudinal direction from AP and
pffiffiffi LAT fields obeysmoverall > 2$Ssetup = P , there is a significant overall displacement at the 95% confidence level and the process should be investigated to find the cause of the error. This may be caused by e.g. a distinct difference between CT and linear accelerator couches [4], or by the precision of laser alignment either of the simulator or the treatment unit, or by a systematic error of the observer [15]. In such cases it may be considered a good practice to increase treatment margins in order to compensate for such an error. The causes of an error should be identified and minimized, but this is beyond the scope of this report. Levene’s test has shown that the overall standard deviations of the displacements increase significantly between daily and weekly measurements in the dorsoventral and caudocranial directions, measured in the lateral PIs (Table 3). The PIs measured in the AP fields in the caudocranial and left-right lateral directions, have shown no significant increase in the overall SD of the displacement due to the frequency of measurements (Table 3). Therefore, in practice, PI should be performed more often in the lateral field. As for the other two directions, the acquisition of PIs could be performed once a week. The most common approach aimed at overcoming in patient setup and organ motion uncertainties is to add margins to the clinical target volume (CTV) to create a PTV according to the ICRU [9]. Safety margins are also sometimes added to the critical structure adjacent to the target in order to provide a safety margin ensuring the structure remains protected. Such regions are called planning risk volumes (PRVs). However, such approach, while producing a treatment that is robust against uncertainties, necessarily increases the radiation exposure of both the healthy tissue and the organ of risk. Secondly, during treatment, if some healthy tissue moves into a higher dose area, severe side effects may occur. Moreover, it is possible to generate regions where PRVs and PTVs overlap, creating a conflict in clinical objectives that must be resolved. And finally, if some malignant cells move into the surrounding low dose area, then the tumor will be underdosed with substantial cold spots [16,17].
Table 2a Population systematic and random errors, and CTV to PTV margins [mm].
Longitudinal CC Lateral LR Vertical DV
Systematic (S)
Random (s)
ICRU 62 (Sqrt (S2 þ s2))
Stroom (2S þ 0.7s)
van Herk (2.5S þ 0.7s)
1.51 1.93 1.42
1.86 1.83 1.77
2.40 2.66 2.27
4.32 5.14 4.08
5.08 6.11 4.79
Table 2b CTV-PTV margins for 3DCRT and IMRT patients evaluated from weekly and daily portal images, respectively. Direction Longitudinal (CC) Lateral (LR) Vertical (DV)
Daily PI Weekly PI Daily PI Weekly PI Daily PI Weekly PI
Ssetup [mm]
sset-up [mm]
ICRU 62 [mm]
Stroom [mm]
van Herk [mm]
1.18 1.59 2.12 1.91 1.66 1.58
1.97 1.73 1.77 1.90 2 1.43
2.30 2.35 2.76 2.69 2.60 2.13
3.74 4.39 5.48 5.15 4.72 4.16
4.33 5.19 6.54 6.11 5.55 4.95
536
B. Strbac, V.S. Jokic / Physica Medica 29 (2013) 531e536
Table 3 Population systematic (S) and random errors (s) from some other studies with respective probabilities of target volume coverage [10]. Study
S [mm]
s [mm]
Displacements of errors
Hess [18]
Not reported
Not reported
Bentel [19] Gilbeau [20] De Boer [8] Humphreys [21]
Not reported 1e2.2 1.5e2.0 0.02e0.9
Not reported 0.7e2.3 1.2e2.0 0.4e0.7
Zhang [22] Suzuki [23] (SUZUKI Minoru, 2006) Gupta [10] Present study
1.5e3.2 0.7e1.3
1.1e2.9 0.7e1.6
0.96e1.2 1.42e1.93
1.94e2.48 1.77e1.86
3 mm for 50% of coverage 9 mm for 95% of coverage 5e10 mm (87e90% with 5 mm margin 4.5e5.5 mm for 90% probability of target coverage e 3 mm for 95% of the errors 5 mm for 99% of errors 5.5 mm for 90% probability of target coverage 5 mm margin for PTV and 3 mm for PRV. Probability values not specified. <5 mm margin in all directions 6.1 mm CTV-PTV LR direction, 5.1 mm CTV-PTV CC direction, 4.8 mm CTV-PTV DV direction.
The range of the errors we reported is comparable to the analogous results of other studies given in Table 3. In those studies, standard deviations of the systematic and random set-up errors varied between 0.02e3.2 mm and 0.4e2.9 mm respectively, whereas CTV-PTV margin ranged from 3 mm to 9 mm [18,21]. Conclusion In our institution margin approach is used to improve patient positioning and perform regular position correction. Smaller PTV margins will result in a lower normal tissue complication probability. PTV margins calculated according to ICRU, Stroom and van Herk models (Table 2) range between 1.95 and 6.16 mm in the three directions. Based on our results we can conclude that a 6-mm extension of CTV to PTV margin, as the lower limit, can be enough to ensure that 90% of patients in the population receive a minimum cumulative CTV dose of at least 95% of the prescribed dose. Further reduction in margins may be considered if an adequate correction strategy is applied. However, set-up errors still remain one of the several sources of uncertainty. In order to fully use the potential of IMRT and 3DCRT in complex cases involving considerable organ motion, we believe that the use of the new image-guided radiotherapy modalities would allow for a decrease in PTV margins. References [1] Thomas J, Burton K, Jefferies S, Burnet G. Defining the tumour and target volumes for radiotherapy. Cancer Imaging 2004;4(2):153e61. [2] Lovelock DM, Yorke ED, Kriminski S, Lee N, Amols HI, Kang H. Accurate positioning for head and neck cancer patients using 2D and 3D image guidance. J Appl Clin Med Phys 2010;12(1):3270. [3] Vight Lisette van der, Huizenga Henk, Kaanders Johannes HAM, Visser Andries G, Emile NJ Th, Lin van. Set-up improvement in head and neck radiotherapy using a 3D off-line EPID-based correction protocol and a customised head and neck support. Radiother Oncol 2003;68:137e48. [4] Coffey M, Greener T, Hall C, Van Herk M, Mijnheer B, Harrison A, et al. Geometric uncertainties in radiotherapy: technical overview of geometric uncertainties in radiotherapy. BIR Working Party 2003. [5] Wu Qiuwen, Worthy Danielle. Dosimetric assessment of rigid setup error by CBCT for HN-IMRT. J Appl Clin Med Phys 2010;11(3). [6] Wolfsberger Luciant, Allen Aaron M, James Steven, Tishler Roy B, Court Laurence E. Clinical experience of the importance of daily portal imaging for head and neck IMRT treatments. J Appl Clin Med Phys 2008;9(3):2756. [7] Lebesque JV, Hart AA, Vijlbrief RE, Bijhold J. Maximizing setup accuracy using portal images as applied to a conformal boost technique for prostatic cancer. Radiother Oncol 1992;24(4):261e71.
[8] Van Sörnsen De Koste JR, Creutzberg CL, Visser AG, Levendag PC, Heijmen BJ, De Boer HC. Electronic portal image assisted reduction of systematic set-up errors in head and neck irradiation. Radiother Oncol 2001;61(3):299e308. [9] International Commission on Radiation Units and Measurements, ICRU-62. Prescribing, recording and reporting photon beam therapy (supplement to ICRU 50). Bethesda, MD: ICRU; 1999. [10] Chopra Supriya, Kadam Avinash, Agarwal Jai P, Reena Devi P, GhoshLaskar Sarbani, Dinshaw Ketayun A, et al. Assessment of three-dimensional set-up errors in conventional head and neck radiotherapy using electronic portal imaging device. Radiat Oncol 2007;2(44). [11] Heijmen BJM, Stroom JC. Geometrical uncertainties, radiotherapy planning margins, and the ICRU-62 report. Radiother Oncol 2002;64:75e83. [12] Remeijer P, Rasch C, Lebesque JV, van Herk M. The probability of correct target dosage: dose-population histograms for deriving treatment margins in radiotherapy. Int J Radiat Oncol Biol Phys 2000;47(4):1121e35. [13] Mercke Claes, Johansson Karl-Axel, Samuelsson Anna. Systematic set-up errors for IMRT in the head and neck region: effect on dose distribution. Radiother Oncol 2003;66:303e11. [14] Lin Z, Donaldson SS, Le QT, Tate D, Goffinet DR, Wolden S, et al. Dosimetric effects of patient displacement and collimator and gantry angle misalignment on intensity modulated radiation therapy. Radiother Oncol 2000;56(1):97e 108. [15] Pichenot Charlotte, Castaing Marine, Auperin Anne, Lefkopoulos Dimitri, Arriagada Rodrigo, Bourhis Jean, et al. Interfractional set-up errors evaluation by daily electronic portal imaging of IMRT in head and neck cancer patients. Acta Oncol Stockholm Sweden 2009;48(3):440e5. [16] Bortfeld Thomas, Tsitsiklis John N, Chan Timothy CY. A robust approach to IMRT optimization. Phys Med Biol 2006;51:2567e83. [17] Zinchenko Yuriy, Henderson Shane G, Sharpez Michael B, Chu Millie. Robust optimization for intensity modulated radiation therapy treatment planning under uncertainty. Phys Med Biol 2005;50:5463. [18] Kortmann RD, Jany R, Hamberger A, Bamberg M, Hess CF. Accuracy of field alignment in radiotherapy of head and neck cancer utilizing individualized face mask immobilization: a retrospective analysis of clinical practice. Radiother Oncol 1995;34(1):69e72. [19] Marks LB, Hendren K, Brizel DM, Bentel GC. Comparison of two head and neck immobilization systems. Int J Radiat Oncol Biol Phys 1997;38(4):867e73. [20] Octave-Prignot M, Loncol T, Renard L, Scalliet P, Grégoire V, Gilbeau L. Comparison of setup accuracy of three different thermoplastic masks for the treatment of brain and head and neck tumors. Radiother Oncol 2001;58(2):155e62. [21] Guerrero Urbano MT, Mubata C, Miles E, Harrington KJ, Bidmead M, Nutting CM, et al. Assessment of a customised immobilisation system for head and neck IMRT using electronic portal imaging. Radiother Oncol 2005;77(1): 39e44. [22] Garden AS, Lo J, Ang KK, Ahamad A, Morrison WH, Rosenthal DI, et al. Multiple regions-of-interest analysis of setup uncertainties for head-and-neck cancer radiotherapy. Int J Radiat Oncol Biol Phys 2006;64(5):1559e69. [23] Minoru Suzuki, Yasumasa Nishimura, Kiyoshi Nakamatsu, Masahiko Okumura, Hisayuki Hashiba, Ryuta Koike, Shuichi Kanamori, et al. Analysis of interfractional set-up errors and intrafractional organ motions during IMRT for head and neck tumors to define an appropriate planning target volume (PTV)- and planning organs at risk volume (PRV)-margins. Radiother Oncol 2006;78(3): 283e90. [24] McKenzie A, van Herk M, Mijnheer B. Margins for geometric uncertainty around organs at risk in radiotherapy. Radiother Oncol 2002;62:299e307.
Contents lists available at SciVerse ScienceDirect
Physica Medica journal homepage: http://www.physicamedica.com
Original paper
Evaluation of set-up errors in head and neck radiotherapy using electronic portal imaging Bojan Strbac a, *, Vesna Spasic Jokic b,1 a b
International Medical Centres, Centre for Radiotherapy, Dvanaest beba bb, 78000 Banja Luka, Bosnia and Herzegovina University of Novi Sad, Faculty of Technical Science, Trg Dositeja Obradovica 6, 21000 Novi Sad, Serbia
a r t i c l e i n f o
a b s t r a c t
Article history: Received 20 April 2012 Received in revised form 28 November 2012 Accepted 7 December 2012 Available online 2 January 2013
Introduction: The aim of this study was to evaluate three-dimensional (3D) set-up errors and propose optimum margins for planning target volume (PTV) coverage in head and neck radiotherapy. Methods: Thirty-five patients were included in the study. The total number of portal images studied was 632. Population systematic (S) and random (s) errors for the patients with head and neck cancer were evaluated based on the portal images in the caudocranial longitudinal (CC) and left-right lateral (LR) direction measured in the anterior-posterior (AP) field, as well as from the images in the caudocranial longitudinal (CC) and dorsoventral lateral (DV) direction measured in the lateral (LAT) field. The values for the clinical-to-planning target volume (CTV-PTV) margins were calculated using ICRU Report 62 recommendations, along with Stroom’s and van Herk’s formulae. Results: The standard deviations of systematic set-up errors (S) ranged from 1.51 to 1.93 mm while the standard deviations of random set-up (s) errors fell in between 1.77 and 1.86 mm. The mean 3D vector length of displacement was 2.66 mm. PTV margins calculated according to ICRU, Stroom’s and van Herk’s models were comprised between 1.95 and 6.16 mm in the three acquisition directions. Discussion and conclusions: Based on our results we can conclude that a 6-mm extension of CTV to PTV margin, as the lower limit, is enough to ensure that 90% of the patients treated for head and neck cancer will receive a minimum cumulative CTV dose greater than or equal to 95% of the prescribed dose. Ó 2012 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.
Keywords: Set-up errors Radiotherapy 3DCRT IMRT Head and neck Portal images PTV margins
Introduction Cancer is the second leading cause of death in the industrialized countries and the only major disease for which death rates are increasing. The demand for cancer care is expected to increase over the coming decade. Radiotherapy plays a key role in cancer treatment. According to the statistical data from the Republic of Serbia and the Republic of Srpska (a part of Bosnia and Herzegovina), about 0.5% of the population will develop a new cancer per year and at least a half of them will require radiotherapy at some stage of their illness. Among the patients having radiotherapy, more than 65% are treated with curative intent, often in combination with surgery and chemotherapy. In addition, radiotherapy has an important role in the palliation of cancer symptoms.
* Corresponding author. Tel. þ387 65859068. E-mail addresses: [email protected], [email protected] (B. Strbac), [email protected] (V.S. Jokic). 1 Tel.: þ381 21 485 2569.
Radiotherapy treatment is a local one so the main goal of radiation therapy planning is to maximize the dose to the target while limiting the dose to nearby healthy organs (“risk organs”), in order to improve the control of tumor growth and to reduce side effects. Evaluation of treatment-related errors, especially those resulting from patient set-up and organ motion, is an increasingly important part of the clinical radiotherapy process. The linear accelerator with multileaf collimation technology has evolved to such an extent that radiation dose can be delivered with high accuracy to volumes in regular objects. Unfortunately, the accuracy of dose delivery to dynamic cancer targets is limited by the uncertainty in many of the treatment parameters. A treatment uncertainty usually includes: an uncertainty in organ shape and motion, beam geometry and patient set-up. Inter-fractional (day-to-day) geometric change occurs over the weeks of therapy, due to digestive processes, change in breathing patterns, differences in patient set-up, and treatment responses such as the growth or the shrinkage of the tumor or the nearby risk organs (e.g., the parotids in head and neck treatment). These uncertainties are taken into account by population-based “uncertainty” margins around the target area, which may be excessive or conservative and are applied to the structures identified
1120-1797/$ e see front matter Ó 2012 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejmp.2012.12.001
532
B. Strbac, V.S. Jokic / Physica Medica 29 (2013) 531e536
before initiation of the therapy [4]. A reduction in treatment set-up errors can be achieved by improving laser alignment, table and gantry stability and accuracy, and the patient set-up procedures by the use of appropriate immobilization methods [3]. Each set-up error consists of a systematic component, i.e. the same deviation in the same direction for each fraction throughout the whole course of treatment, as well as a random component, varying from day-to-day [4]. Both types of errors were investigated for 35 head and neck radiotherapy (HN-RT) patients by applying an electronic portal imaging device (EPID). All patients included in this report underwent 3D conformal and intensity modulated head and neck radiotherapy (3DCRT, IMRT) in our institute. There are three main types of volume defined in radiotherapy planning. The first one is given by the position and extent of gross tumor, i.e. what can be seen, palpated or imaged, and is defined as the gross tumor volume (GTV). The second volume contains the GTV, plus a margin reflecting sub-clinical disease spread which therefore cannot be fully imaged; this is known as the clinical target volume (CTV) [1]. The CTV is important because this volume must be appropriately treated to achieve cure. The third volume, the planning target volume (PTV), allows for uncertainties in planning or treatment delivery. It is a geometric concept designed to ensure that the radiotherapy dose is actually delivered to the CTV [1]. Radiotherapy planning must always consider critical normal tissue structures, known as organs at risk (ORs). In some specific circumstances, it is necessary to add a margin analogous to the PTV margin around an OR in order to ensure that the organ cannot receive a higher-than-safe dose; this gives a planning organ at risk volume [24]. Accurate positioning in the radiation treatment is important because set-up errors can result in significant underdose to the tumor and/or overdose to one or more critical organs [2]. The first aim of this study was to estimate systematic and random set-up uncertainties by means of portal imaging for patients with head and neck cancer treated with 3DCRT and IMRT technique, and to find optimal CTV to PTV margin expansion. The second aim of the study was to estimate the impact of the frequency of online verification on set-up accuracy and set-up margins. Materials and methods Thirty-five patients were included in this study. Twenty-nine of them underwent 3D conformal radiotherapy of head and neck region, with at least one weekly image (patients ¼ 29) whereas the remaining six underwent IMRT with daily imaging (patients ¼ 6). HN-RT patients were immobilized using 5-point Orfit masks (four fixation points plus an additional one on the top of the head). After a CT scan of the patient was obtained, DRRs of the beam’s eye view (BEV) were computed and stored in the treatment planning
system (TPS) and were successively considered to be the reference images. The DRR images were saved in a database accessible to our software. In our clinic we use ARIA (Varian Medical System) as the information and image management system that aggregates patient data into a fully-electronic medical chart. Portal images were acquired by using the EPID system attached to Varian linear accelerators. To ensure accuracy on the treatment couch, patient positioning adjustments were made aligning surface markers (i.e. tattoos) placed at the time of simulation. Radiation therapists (four of them) visually compared the portal images to planned DRRs by manually matching bone anatomy landmarks by means of anatomy matching software PortalVisionÔ Advanced Imaging (Varian Medical System). The therapists had a full control over image display contrast, window level, image filter and magnification. For patients who were treated by 3DCRT, the measurement of set-up errors was performed at least three times during the first week of treatment and on a weekly basis thereafter. Portal imaging was done before the treatment and set-up errors were assessed. The offline correction protocol was created so as to correct set-up errors should the mismatch have been found to be more than 3 mm in any direction. No action was taken if the mismatch was within 3 mm in all direction. If the PIs were incorrect, the patient position was corrected within the mask and the additional PI was performed to recheck the positioning [6]. The corrective shifts (in table parameters) were applied and kept constant every day for subsequent treatment until the next imaging was done, typically the following week. If there were more than one portal imaging in a day, the setup errors were analyzed from the first and last portal of the day. Compared to 3DCRT radiotherapy, sharp dose gradients are believed to make Intensity Modulated Radiotherapy (IMRT) dose distribution more sensitive to geometric uncertainties [5]. This is a major concern, as the delivered dose to the patient may deviate significantly from the approved plan, not only preventing the benefits of IMRT over 3DCRT from being clinically realized, but also potentially resulting in the delivery of a worse treatment. For all these reasons, all the head and neck IMRT patients at our clinic were treated using an image-guided radiation therapy technique based on daily orthogonal portal images. Set-up deviations were measured in 318 anterior and 314 lateral portal images and were assessed in caudocranial longitudinal, and in left-right lateral direction measured in AP field as well as in dorsoventral vertical, and caudocranial longitudinal direction measured in the lateral field. Displacements, i.e. total set-up error m, were assessed offline with Varian Offline Review anatomy matching software (Fig. 1). Mean displacements, population systematic and random errors, and 3D vector of displacement were calculated.
Figure 1. Varian offline review (a) AP images; (b) Lateral images (portal and DRR).
B. Strbac, V.S. Jokic / Physica Medica 29 (2013) 531e536
There is a time gap between PIs acquired for 3DCRT and those acquired for IMRT treatment, because our institution has only recently started with IMRT treatment of patients. This is the main reason why only six patients included in this study underwent IMRT treatment. Patient set-up analysis The mismatch result, i.e. set-up deviation m, is a deviation combining both random and systematic component. For each patient, systematic and random deviations were calculated for all the three directions separately. A patient set-up deviation is defined as the difference between the actual and intended positions of the part of the patient’s body irradiated. The standard deviation of the random errors is denoted as s, defined as the standard deviation of the day-to-day set-up positions averaged over all the patients in the group. The distribution of systematic P deviations is denoted as , defined as the standard deviation of the distribution of average set-up deviations per patient [7]. The reliability of the following statistical approach which estimates these standard deviations depends on the total number of patients P and images N used in the study. All the patients are assumed to be consistent in terms of set-up technique. The calculation method assumes that both random and systematic components are normally distributed [4]. The mean deviation of np images (measurements) i.e. the individual systematic set-up deviation for a patient p is given by:
mp ¼
533
1 X m : np i ¼ 1;n ðPIDRRÞ;i
(1)
p
The individual random set-up deviation, i.e. the standard deviation of the distribution of m deviations around mp for a patient p in a given direction (inter-fractional):
srand;p
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 2 u1 X mðPIDRRÞ;i mp : ¼ t np i ¼ 1;n
(2)
p
The overall mean systematic error for all the P patients in the study:
moverall ¼
1 X np $mp ; N p ¼ 1;P
(3)
where N is the total number of images in the study. The Equation (3) is weighed according to the number of the portal images taken per patient. The random set-up error for all the P patients in a given direction i.e. the standard deviation of the srand,p distribution:
ssetup
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u X u P s2 ¼ t np 1 : N P p ¼ 1;P rand;p
(4)
Figure 2. Distribution of total deviations m, measured in the (a) caudocranial longitudinal and (b) left-right lateral direction from the anterior-posterior field (AP) as well as in the (c) caudocranial longitudinal direction and (d) dorsoventral vertical direction from lateral field (LAT).
534
B. Strbac, V.S. Jokic / Physica Medica 29 (2013) 531e536
Figure 3. Scatter plots illustrate 2D displacement vectors by crosses in the AP field and the lateral field respectively. The displacements are expressed in mm for all the patients in the study.
The standard deviation of the random treatment set-up error for all theP patients in a given direction i.e. the standard deviation for the mp distribution is as follows:
Ssetup
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u X 2 u 1 ¼ t np mp moverall : NðP 1Þ p ¼ 1;P
(5)
pffiffiffi Values of moverall greater than t$Ssetup = P would indicate a statistically significant overall systematic error at the 95% confidence limit and should be investigated (the mean deviations mp were assumed to follow a t-distribution; for a 95% confidence level and P 1 ¼ 34 degrees of freedom the tabulated t value is z 2) [4]. To calculate a 3D vector length, the measured 1D deviations in the three directions are combined quadratically. The mean 3D set-up deviations (one value per patient) are calculated as the average length of the 3D vector deviations over all measured fractions [3]. Due to a limited number of observed fractions per patient the valueSsetup will tend to be overestimated due to the presence of the random errors. De Boer et al. [8] provide a correction that can be applied to give an unbiased estimator of variance:
ðSunbiased Þ2 ¼ ðSbias Þ2
2 1 X sinter;p P P ¼ 1;P np
(6)
The above method assumes the normal distribution of random and systematic components [4]. Several mathematic formulae have been recommended for generating CTV-PTV margins. The ICRU 62 [9] states that systematic and random uncertainties should be added in quadrature, which should then be used for margin calculation. However, this approach assumes that random and systematic errors have an equal contribution to the dose distribution, which might not be necessarily true. Random errors blur the dose distribution whereas systematic errors cause a shift in the cumulative dose distribution [10]. Using coverage probability matrices and dose population histograms, Heijmen and Stroom [11] and van Herk et al. [12] have suggested formulae incorporating these differential effects. Stroom’s margin recipe ð2S þ 0:7sÞ ensures that, on average, 99% of the CTV receives more than or equal to 95% of the prescribed dose. The formula by van Herk ð2:5S þ 0:7sÞ ensures that 90% of patients in the population receive a minimum cumulative CTV dose of at least 95% of the prescribed dose [12]. To calculate a 3D vector length, the measured 1D deviations in the three directions are combined quadratically [3].
Results The overall mean systematic deviation moverall was calculated according to Equation (3). The random set-up error,ssetup i.e. the standard deviation of the srand;p for all the patients, was calculated according to Equation (4), and the systematic error Ssetup i.e. the standard deviation of mp , was calculated according to Equation (5) Fig. 2. Figure 3 shows the total deviations m, measured in the (a) caudocranial longitudinal and (b) left-right lateral direction from the anterior-posterior field (AP) along with those measured in the (c) caudocranial longitudinal direction and (d) dorsoventral vertical direction from the lateral field (LAT). The distribution appears to be a Gaussian one. Figure 3 is a graphic representation of the displacement vectors in the AP and LAT field, respectively. The results summarized in Table 1a take into account all the data acquired from daily and weekly PIs. In Table 1b an analysis of the distributions of daily and weekly acquired data is shown. The differences in standard deviations (SDs) of the overall displacement between daily and weekly portals were tested using Levene’s test for homogeneity of SDs [13] (Berrin Pehlivan, 2009). A p-value of <0.05 was considered statistically significant. The mean 3D vector length of displacement was 2.66 mm. Results for CTV-PTV margins were calculated using ICRU Report 62 as well as Stroom’s and van Herk’s formulae [9,11,12] (Table 2). Discussion The purpose of this study was to evaluate the patient set-up errors for patients with head and neck cancer. The sensitivity of a treatment plan to set-up errors is dependent on the sharpness of
Table 1a The summary of results for the population systematic S and random s errors in the patients with head and neck cancer evaluated from portal images in the caudocranial longitudinal and left-right lateral direction measured in the AP field, and dorsoventral and caudocranial field measured in the LAT field. Field
AP
Direction
Caudocranial longitudinal
Left-right lateral
LAT Dorsoventral vertical
Caudocranial longitudinal
Min deviation [mm] Max deviation [mm] moverall [mm] sset-up [mm] Ssetup [mm]
6 10 1.55 1.86 1.51
9 6 0.01 1.83 1.93
9 7 0.15 1.77 1.42
5 8 1.09 1.81 1.52
B. Strbac, V.S. Jokic / Physica Medica 29 (2013) 531e536
535
Table 1b The summaries of results for the population systematic S and random s errors for 3DCRT patients evaluated in basis of the weekly portal images and IMRT patients evaluated in basis of the daily portal images. Field
AP
Direction
Caudocranial longitudinal
Left-right lateral
Dorsoventral vertical
Caudocranial longitudinal
2.22 2.21 0.778
2.62 2.54 0.507
1.99 2.41 0.022
2.49 1.80 0.000
Daily PI Weekly PI
Overall standard deviation [mm] Levene’s p value
Lateral
the gradients in the plan. Increasing the number of beams in an IMRT treatment plan will increase the sharpness of the gradients around OARs and consequently the sensitivity to set-up errors increase [13]. Xing et al. [14] observed that a 3-mm error in the couch location in the AP direction resulted in a 38% decrease of the minimal target dose or in a 41% increase of the minimal spinal cord dose. Therefore, it is highly important to quantify and reduce patient set-up errors. In our institution we have decided to perform daily online verification in all the patients treated with IMRT and in the patients treated with 3DCRT where critical organs at risk are located in close proximity of the PTV. For patients where the prescribed dose does not exceed tolerance doses for organs at risk, the frequency of PIs is lower (weekly). We analyzed the patient set-up accuracy using the concept of systematic and random errors. The systematic component of any error can be defined as a deviation that occurs in the same direction and is of a similar magnitude for each fraction throughout the treatment course (treatment preparation errors), and the random component as a deviation that can vary in direction and magnitude for each delivered treatment fraction (treatment execution errors). The differentiation between systematic and random errors is important for the derivation of appropriate safety margins. There are a few limitations in our study, firstly concerning possible rotational errors because portal imaging in the AP and lateral projection has not captured rotational shifts. Secondly, portal images do not give information about organ motion; therefore, we have not accounted for these errors in calculating PTV margins. Suzuki et al. [23] reported that systematic and random set-up errors for organ motion ranged from 0.2 to 0.8 mm and 0.3e0.6 mm, respectively. Thirdly, a potential bias in the interpretation of the data concerning set-up errors lies in the observer’s judgment during image registration. Van Lin et al. [3] reported that the difference between two observers averaged over all patients is 0.1 mm or less for all directions. Since the absolute values of overall mean systematic error for patient set-up deviation in the longitudinal direction from AP and
pffiffiffi LAT fields obeysmoverall > 2$Ssetup = P , there is a significant overall displacement at the 95% confidence level and the process should be investigated to find the cause of the error. This may be caused by e.g. a distinct difference between CT and linear accelerator couches [4], or by the precision of laser alignment either of the simulator or the treatment unit, or by a systematic error of the observer [15]. In such cases it may be considered a good practice to increase treatment margins in order to compensate for such an error. The causes of an error should be identified and minimized, but this is beyond the scope of this report. Levene’s test has shown that the overall standard deviations of the displacements increase significantly between daily and weekly measurements in the dorsoventral and caudocranial directions, measured in the lateral PIs (Table 3). The PIs measured in the AP fields in the caudocranial and left-right lateral directions, have shown no significant increase in the overall SD of the displacement due to the frequency of measurements (Table 3). Therefore, in practice, PI should be performed more often in the lateral field. As for the other two directions, the acquisition of PIs could be performed once a week. The most common approach aimed at overcoming in patient setup and organ motion uncertainties is to add margins to the clinical target volume (CTV) to create a PTV according to the ICRU [9]. Safety margins are also sometimes added to the critical structure adjacent to the target in order to provide a safety margin ensuring the structure remains protected. Such regions are called planning risk volumes (PRVs). However, such approach, while producing a treatment that is robust against uncertainties, necessarily increases the radiation exposure of both the healthy tissue and the organ of risk. Secondly, during treatment, if some healthy tissue moves into a higher dose area, severe side effects may occur. Moreover, it is possible to generate regions where PRVs and PTVs overlap, creating a conflict in clinical objectives that must be resolved. And finally, if some malignant cells move into the surrounding low dose area, then the tumor will be underdosed with substantial cold spots [16,17].
Table 2a Population systematic and random errors, and CTV to PTV margins [mm].
Longitudinal CC Lateral LR Vertical DV
Systematic (S)
Random (s)
ICRU 62 (Sqrt (S2 þ s2))
Stroom (2S þ 0.7s)
van Herk (2.5S þ 0.7s)
1.51 1.93 1.42
1.86 1.83 1.77
2.40 2.66 2.27
4.32 5.14 4.08
5.08 6.11 4.79
Table 2b CTV-PTV margins for 3DCRT and IMRT patients evaluated from weekly and daily portal images, respectively. Direction Longitudinal (CC) Lateral (LR) Vertical (DV)
Daily PI Weekly PI Daily PI Weekly PI Daily PI Weekly PI
Ssetup [mm]
sset-up [mm]
ICRU 62 [mm]
Stroom [mm]
van Herk [mm]
1.18 1.59 2.12 1.91 1.66 1.58
1.97 1.73 1.77 1.90 2 1.43
2.30 2.35 2.76 2.69 2.60 2.13
3.74 4.39 5.48 5.15 4.72 4.16
4.33 5.19 6.54 6.11 5.55 4.95
536
B. Strbac, V.S. Jokic / Physica Medica 29 (2013) 531e536
Table 3 Population systematic (S) and random errors (s) from some other studies with respective probabilities of target volume coverage [10]. Study
S [mm]
s [mm]
Displacements of errors
Hess [18]
Not reported
Not reported
Bentel [19] Gilbeau [20] De Boer [8] Humphreys [21]
Not reported 1e2.2 1.5e2.0 0.02e0.9
Not reported 0.7e2.3 1.2e2.0 0.4e0.7
Zhang [22] Suzuki [23] (SUZUKI Minoru, 2006) Gupta [10] Present study
1.5e3.2 0.7e1.3
1.1e2.9 0.7e1.6
0.96e1.2 1.42e1.93
1.94e2.48 1.77e1.86
3 mm for 50% of coverage 9 mm for 95% of coverage 5e10 mm (87e90% with 5 mm margin 4.5e5.5 mm for 90% probability of target coverage e 3 mm for 95% of the errors 5 mm for 99% of errors 5.5 mm for 90% probability of target coverage 5 mm margin for PTV and 3 mm for PRV. Probability values not specified. <5 mm margin in all directions 6.1 mm CTV-PTV LR direction, 5.1 mm CTV-PTV CC direction, 4.8 mm CTV-PTV DV direction.
The range of the errors we reported is comparable to the analogous results of other studies given in Table 3. In those studies, standard deviations of the systematic and random set-up errors varied between 0.02e3.2 mm and 0.4e2.9 mm respectively, whereas CTV-PTV margin ranged from 3 mm to 9 mm [18,21]. Conclusion In our institution margin approach is used to improve patient positioning and perform regular position correction. Smaller PTV margins will result in a lower normal tissue complication probability. PTV margins calculated according to ICRU, Stroom and van Herk models (Table 2) range between 1.95 and 6.16 mm in the three directions. Based on our results we can conclude that a 6-mm extension of CTV to PTV margin, as the lower limit, can be enough to ensure that 90% of patients in the population receive a minimum cumulative CTV dose of at least 95% of the prescribed dose. Further reduction in margins may be considered if an adequate correction strategy is applied. However, set-up errors still remain one of the several sources of uncertainty. In order to fully use the potential of IMRT and 3DCRT in complex cases involving considerable organ motion, we believe that the use of the new image-guided radiotherapy modalities would allow for a decrease in PTV margins. References [1] Thomas J, Burton K, Jefferies S, Burnet G. Defining the tumour and target volumes for radiotherapy. Cancer Imaging 2004;4(2):153e61. [2] Lovelock DM, Yorke ED, Kriminski S, Lee N, Amols HI, Kang H. Accurate positioning for head and neck cancer patients using 2D and 3D image guidance. J Appl Clin Med Phys 2010;12(1):3270. [3] Vight Lisette van der, Huizenga Henk, Kaanders Johannes HAM, Visser Andries G, Emile NJ Th, Lin van. Set-up improvement in head and neck radiotherapy using a 3D off-line EPID-based correction protocol and a customised head and neck support. Radiother Oncol 2003;68:137e48. [4] Coffey M, Greener T, Hall C, Van Herk M, Mijnheer B, Harrison A, et al. Geometric uncertainties in radiotherapy: technical overview of geometric uncertainties in radiotherapy. BIR Working Party 2003. [5] Wu Qiuwen, Worthy Danielle. Dosimetric assessment of rigid setup error by CBCT for HN-IMRT. J Appl Clin Med Phys 2010;11(3). [6] Wolfsberger Luciant, Allen Aaron M, James Steven, Tishler Roy B, Court Laurence E. Clinical experience of the importance of daily portal imaging for head and neck IMRT treatments. J Appl Clin Med Phys 2008;9(3):2756. [7] Lebesque JV, Hart AA, Vijlbrief RE, Bijhold J. Maximizing setup accuracy using portal images as applied to a conformal boost technique for prostatic cancer. Radiother Oncol 1992;24(4):261e71.
[8] Van Sörnsen De Koste JR, Creutzberg CL, Visser AG, Levendag PC, Heijmen BJ, De Boer HC. Electronic portal image assisted reduction of systematic set-up errors in head and neck irradiation. Radiother Oncol 2001;61(3):299e308. [9] International Commission on Radiation Units and Measurements, ICRU-62. Prescribing, recording and reporting photon beam therapy (supplement to ICRU 50). Bethesda, MD: ICRU; 1999. [10] Chopra Supriya, Kadam Avinash, Agarwal Jai P, Reena Devi P, GhoshLaskar Sarbani, Dinshaw Ketayun A, et al. Assessment of three-dimensional set-up errors in conventional head and neck radiotherapy using electronic portal imaging device. Radiat Oncol 2007;2(44). [11] Heijmen BJM, Stroom JC. Geometrical uncertainties, radiotherapy planning margins, and the ICRU-62 report. Radiother Oncol 2002;64:75e83. [12] Remeijer P, Rasch C, Lebesque JV, van Herk M. The probability of correct target dosage: dose-population histograms for deriving treatment margins in radiotherapy. Int J Radiat Oncol Biol Phys 2000;47(4):1121e35. [13] Mercke Claes, Johansson Karl-Axel, Samuelsson Anna. Systematic set-up errors for IMRT in the head and neck region: effect on dose distribution. Radiother Oncol 2003;66:303e11. [14] Lin Z, Donaldson SS, Le QT, Tate D, Goffinet DR, Wolden S, et al. Dosimetric effects of patient displacement and collimator and gantry angle misalignment on intensity modulated radiation therapy. Radiother Oncol 2000;56(1):97e 108. [15] Pichenot Charlotte, Castaing Marine, Auperin Anne, Lefkopoulos Dimitri, Arriagada Rodrigo, Bourhis Jean, et al. Interfractional set-up errors evaluation by daily electronic portal imaging of IMRT in head and neck cancer patients. Acta Oncol Stockholm Sweden 2009;48(3):440e5. [16] Bortfeld Thomas, Tsitsiklis John N, Chan Timothy CY. A robust approach to IMRT optimization. Phys Med Biol 2006;51:2567e83. [17] Zinchenko Yuriy, Henderson Shane G, Sharpez Michael B, Chu Millie. Robust optimization for intensity modulated radiation therapy treatment planning under uncertainty. Phys Med Biol 2005;50:5463. [18] Kortmann RD, Jany R, Hamberger A, Bamberg M, Hess CF. Accuracy of field alignment in radiotherapy of head and neck cancer utilizing individualized face mask immobilization: a retrospective analysis of clinical practice. Radiother Oncol 1995;34(1):69e72. [19] Marks LB, Hendren K, Brizel DM, Bentel GC. Comparison of two head and neck immobilization systems. Int J Radiat Oncol Biol Phys 1997;38(4):867e73. [20] Octave-Prignot M, Loncol T, Renard L, Scalliet P, Grégoire V, Gilbeau L. Comparison of setup accuracy of three different thermoplastic masks for the treatment of brain and head and neck tumors. Radiother Oncol 2001;58(2):155e62. [21] Guerrero Urbano MT, Mubata C, Miles E, Harrington KJ, Bidmead M, Nutting CM, et al. Assessment of a customised immobilisation system for head and neck IMRT using electronic portal imaging. Radiother Oncol 2005;77(1): 39e44. [22] Garden AS, Lo J, Ang KK, Ahamad A, Morrison WH, Rosenthal DI, et al. Multiple regions-of-interest analysis of setup uncertainties for head-and-neck cancer radiotherapy. Int J Radiat Oncol Biol Phys 2006;64(5):1559e69. [23] Minoru Suzuki, Yasumasa Nishimura, Kiyoshi Nakamatsu, Masahiko Okumura, Hisayuki Hashiba, Ryuta Koike, Shuichi Kanamori, et al. Analysis of interfractional set-up errors and intrafractional organ motions during IMRT for head and neck tumors to define an appropriate planning target volume (PTV)- and planning organs at risk volume (PRV)-margins. Radiother Oncol 2006;78(3): 283e90. [24] McKenzie A, van Herk M, Mijnheer B. Margins for geometric uncertainty around organs at risk in radiotherapy. Radiother Oncol 2002;62:299e307.