Home

Search

Collections

Journals

About

Contact us

My IOPscience

A simulation study of irregular respiratory motion and its dosimetric impact on lung tumors

This article has been downloaded from IOPscience. Please scroll down to see the full text article. 2011 Phys. Med. Biol. 56 845 (http://iopscience.iop.org/0031-9155/56/3/019) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 130.88.240.14 The article was downloaded on 01/08/2011 at 16:00

Please note that terms and conditions apply.

IOP PUBLISHING Phys. Med. Biol. 56 (2011) 845859

PHYSICS IN MEDICINE AND BIOLOGY

doi:10.1088/0031-9155/56/3/019

A simulation study of irregular respiratory motion and its dosimetric impact on lung tumors
Y D Mutaf1 , C J Scicutella, D Michalski, K Fallon, E D Brandner, G Bednarz and M S Huq
Department of Radiation Oncology, University of Pittsburgh Medical Center, Pittsburgh, PA 15232, USA E-mail: ydm@mutaf.org

Received 5 August 2010, in nal form 29 November 2010 Published 17 January 2011 Online at stacks.iop.org/PMB/56/845 Abstract This study is aimed at providing a dosimetric evaluation of the irregular motion of lung tumors due to variations in patients respiration. Twenty-three lung cancer patients are retrospectively enrolled in this study. The motion of the patient clinical target volume is simulated and two types of irregularities are dened: characteristic and uncharacteristic motions. Characteristic irregularities are representative of random uctuations in the observed target motion. Uncharacteristic irregular motion is classied as systematic errors in determination of the target motion during the planning session. Respiratory traces from measurement of patient abdominal motion are also used for the target motion simulations. Characteristic irregular motion was observed to cause minimal changes in target dosimetry with the largest effect of 2.5% 0.9% (1 ) reduction in the minimum target dose (Dmin) observed for targets that move 2 cm on average and exhibiting 50% amplitude variations within a session. However, uncharacteristic irregular motion introduced more drastic changes in the clinical target volume (CTV) dose; 4.1% 1.7% reduction for 1 cm motion and 9.6% 1.7% drop for 2 cm. In simulations with patients abdominal motion, corresponding changes in target dosimetry were observed to be negligible (<0.1%). Only uncharacteristic irregular motion was identied as a clinically signicant source of dosimetric uncertainty. (Some gures in this article are in colour only in the electronic version) 1. Introduction Respiratory motion is a major concern for radiotherapy of lung tumors and requires a thorough assessment of its effects on target localization. With advances in time-resolved imaging
1 Pressent Address: University of Maryland Medical Center, Department of Radiation Oncology, 22 S. Greene St, Baltimore, MD 21201.

0031-9155/11/030845+15$33.00

2011 Institute of Physics and Engineering in Medicine

Printed in the UK

845

846

Y D Mutaf et al

technologies such as 4DCT (Rietzel et al 2005, Mutaf and Brinkmann 2008a), the changes in 3D patient anatomy such as organ motion are measured over time. This additional knowledge is incorporated into the treatment planning process for the patient and used in conjunction with appropriate treatment delivery techniques. In modern radiotherapy, the 3D model of the patient anatomy is valid as long as it is sufciently reproduced between the planning and the actual delivery stages of the radiotherapy. Similarly, the 4D model of the patient used for treatment planning purposes is only deemed valid if the reproducibility of the motion is veried. Analogous to patient setup errors affecting the 3D localization of the target volume, a large component of error in 4D radiotherapy could be the target motion irregularities (Ozhasoglu and Murphy 2002, Chang et al 2007, Juhler Nottrup et al 2007) due to variations in patient respiration. Irregular breathing could be as large as the underlying respiratory motion itself. Respiratory variations could also have intra and inter-fractional components affecting the radiotherapy of mobile tumors in different ways. Using optical measurement methods, Juhler Nottrup et al (2007) evaluated the variations in the chest wall motion of 11 lung cancer patients and found out that inter-fractional variations can be up to ten times larger than the intrafractional variations. Due to observed irregularities in patients respiration, Juhler Nottrup et al (2007) also concluded that margins to account for respiratory motion cannot be determined in one imaging session. Several techniques have been developed to mitigate the occurrence of irregularities in patient respiratory motion. Audio and visual feedback techniques have been demonstrated to be an effective way of achieving a consistent pattern of respiration. George et al (2006) subjected 24 lung cancer patients to audio-visual respiratory feedback techniques and observed signicant improvement in the reproducibility of respiration. Gating technology (Ramsey et al 1999, Wagman et al 2003) is also used to prevent irradiation in the presence of drastic changes in breathing such as sneezing or coughing. Despite the existing studies demonstrating the amount of irregularities in respiratory motion, there is limited information on the dosimetric outcome of irregular breathing and whether or not they have a clinically signicant impact on therapeutic goals of the irradiation. Due to radiation dispersion in tissue, penumbral effects and features of plan topology, geometric observations performed with radiographic imaging or optical methods are not directly translatable to dosimetric conclusions. In the context of patient setup errors, the amount of geometric miss and the associated target margin to ensure sufcient coverage are shown to be proportional but not the same (van Herk 2004). This is a result of differences in the shapes of target denition function perceived visually as a step function at the border of the target and the function of the dosimetric prole across the target with a gradual fall off. Similarly, van Herk et al (2003) analytically demonstrated that the additional target margin required to compensate the motion of the target is substantially smaller than its full motion amplitude. This was also conrmed recently by Mutaf and Brinkmann (2008b) in a more extensive treatment planning study incorporating actual patient data. The aim of the current study is to extend the dosimetric investigation of mobile targets under respiration and evaluate changes in target coverage due to irregularities in its motion.

2. Methods and materials The dosimetric effects of respiratory motion need to be evaluated in relation to the specic treatment delivery techniques used for compensation of the target motion. In this study, target motion was incorporated into treatment planning via an expansion of the target volume by an amount equal to the target excursion during the full respiratory cycle (i.e. no respiratory

Irregular respiratory motion

847

Table 1. Patient target volume statistics and their anatomical location are listed. The number of elds and conformity of the plan dose to the target are also shown in simple and complex geometry plans.

Patient no 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Anatomic location Right lung (LL) Left lung (LL) Right lung (LL) Right lung (LL) Right lung (UL) Right lung (LL) Left lung (LL) Right lung (ML) Right lung (UL) Right lung (ML) Left lung (UL) Right lung (ML) Right lung (LL) Right lung (ML) Left lung (UL) Right lung (UL) Right lung (ML) Left lung (LL) Right lung (UL) Left lung (UL) Right lung (LL) Left lung (UL) Right lung (ML) Average Minimum Maximum

CTV (ml) 68.7 77.4 5.6 58.2 1.4 91.6 36.9 207.6 32.9 3.3 12.0 23.7 5.1 4.9 29.0 6.0 61.8 35.6 1.6 39.2 83.8 20.9 1.8 39.5 1.4 207.6

Simple plans Nelds Conformity 2 3 2 3 3 3 3 3 3 2 3 2 2 3 2 3 2 3 2 2 3 3 3 2.6 2 3 4.62 2.06 2.31 3.45 2.83 4.64 2.30 2.79 2.53 2.32 2.90 9.12 9.72 8.72 8.87 4.45 4.71 2.96 18.00 5.75 2.76 3.03 2.94 4.95 2.06 18.00

Complex plans Nelds Conformity 6 5 7 5 6 8 7 6 5 5 6 6 5 5 6 5 5 6 5 5 7 6 7 5.8 5 8 1.79 1.82 2.40 1.85 2.63 1.85 1.93 1.74 1.98 2.37 1.87 2.02 2.23 2.71 2.00 2.08 1.82 1.93 2.30 1.89 1.65 1.98 2.58 2.06 1.65 2.71

LL: lower lobe; UL: upper lobe; ML: middle lobe.

gating). Irregularities in respiratory motion are investigated using simulations of target motion and analysis of respective clinical treatment plans. 2.1. Patient cohort and dosimetry Twenty-three lung cancer patients previously treated in our clinic are retrospectively analyzed through their informed consent according to an approved University of Pittsburgh institutional review board protocol. Clinical target volumes (CTV) ranged from 1.4 to 207.6 ml. Although, prescription doses varied between 50.0 and 66.0 Gy for actual patient treatments, the prescription doses were all normalized to 60.0 Gy to facilitate easier comparisons for the effects of respiratory motion. A summary of patient information related to treatment plans of the patients is provided in table 1. Due to the radio-sensitivity of lungs to low doses of radiation (Oetzel et al 1995, Graham et al 1999), the clinical treatment plans were designed to minimize irradiation volume and therefore consisted of two or three elds (usually in parallel

848
(b)

Y D Mutaf et al

(a)

Figure 1. Comparison of dose distributions in simple (a) and complex (b) geometry plans in axial and coronal views for patient 12.

opposed and oblique orientations). In this paper, such 3D conformal plans are referred to as simple geometry plans where dose homogeneity within the target ranged between 95% and 115% of the prescription dose. Adopting the RTOG formulation of conformity (Shaw et al 2000) (dened as the ratio of 95% isodose volume to the target volume), the average target conformity in simple plans was 4.95. In addition to simple geometry plans, an alternative treatment plan was also generated incorporating more elds, ranging from ve to eight, in order to produce more conformal target dosimetry. These plans, referred as complex geometry plans, achieved an average conformity of 2.06 in addition to improving dose homogeneity within the target (95% to 105% of prescription dose). The typical dose distributions in simple and complex geometry plans are demonstrated in gure 1 for one of the patients (patient 12). The corresponding dosevolume histograms for target and bi-lateral lungs are also presented in gure 2. In both type of plans, the selection of beams is restricted to the angles orthogonal to the longitudinal axis of the patients (transverse co-planar arrangement). All plans were manually optimized to conform to the shape of the target with static 0.5 cm MLCs and a block margin of 0.8 cm. Hard wedges were utilized for improvement of dose homogeneity. No intensity modulation techniques and optimization engines (i.e. IMRT) are utilized. Plans were further imposed with other constraints such as a maximum tolerable spinal cord dose of 45.0 Gy and a mean heart dose of 40.0 Gy, although the latter constraint was considered on a case-by-case basis depending on the location of the tumor.

Irregular respiratory motion

849

Figure 2. Comparison of target volume and bi-lateral lungs dosevolume histograms for simple (solid lines) and complex (dashed lines) geometry plans shown in gure 1.

2.2. Target selection and contouring Target denitions and contouring adhered to ICRU guidelines (International Commission on Radiation Units and Measurements 1993, 1999). Simulation images were acquired from 4DCT scans using a retrospective phase sorting technique producing ten phase images at equal temporal intervals (i.e. 10% of respiration period). Following institutional protocols, treatment simulation and planning was performed on the end-of-exhalation (EOE) phase dataset extracted from the 4DCT scan (image in which the tumor volume is located most superiorly, typically 50% phase image). The EOE image was preferred for planning since in a typical respiratory cycle, longest amount of time is spent at this phase and therefore produces 3D images with least amount of motion artifacts. CTV was contoured on the EOE images by the patients primary radiation oncologist with consideration for the subclinical extent of the disease. The clinical ITV was nally generated using an asymmetric expansion, internal margin or IM, equal to the amplitude of the target motion. In order to investigate the effects of different amplitudes of motions, the CTV is enlarged by a specic IM amount which is chosen to be equal to the amount of simulated motion and its direction (equation (1)): ITVIM = CTV IM(x, y, z). (1)

In this formalism, ITV volumes are created by adding 1 and 2 cm margins to CTV in the inferior direction as demonstrated in gure 3. These nal structures are referred as ITV1 cm and ITV2 cm within the text, where ITV1 cm = CTV 1 cm and ITV2 cm = CTV 2 cm.. 2.3. Respiratory motion simulations The mechanics of respiration induces an expansioncontraction motion observed mostly in the cranio-caudal (CC) axis of the patient. Several radiological studies (Seppenwoolde et al

850

Y D Mutaf et al

Figure 3. An illustration of the target volumes (CTV and ITV) described in the text. Internal target volumes (ITV) are generated from clinical target volumes (CTV) via application of 1 and 2 cm asymmetric internal margins in the inferior direction.

2002, Murphy et al 2002) have conrmed that largest motion occurs along the CC axis of the patients. Therefore, motion simulations in this study are restricted to the CC axis of the patient where CTV center-of-mass is changed with respect to the amount of simulated motion. Respiratory motion simulations are performed via dose accumulation for voxels within the CTV boundaries through the length of time the respiratory motion assessed. At each dose sampling time, the target center-of-mass was moved with respect to the instantaneous amplitude of the respiratory motion but no deformations were allowed. Therefore, the total number of target voxels and their coordinates within the target reference frame remain the same. Due to the scan resolution of the CT images, target voxels have the dimensions of 1.00 1.00 mm2 in the axial plane and are separated by 1.25 mm slice thickness. Similarly, the simulated motions were also digitized with respect to the scan resolution, i.e. 1.25 mm since the simulated motion is in the longitudinal axis of the patient. In order to improve processing time, the temporal resolution of the simulations are reduced such that the target dose sampling is only performed at 10 equally spaced times within a respiratory cycle (i.e. 10% phase intervals). Two main forms of respiration motion models are used for the simulations: regular and irregular respiration. In a regular respiration study, a single respiratory cycle with a functional form represented in equation (2) was used for target simulations (Lujan et al 2003). A plot of respiratory motion amplitude in this form is shown in gure 4. For simulation of a 1.0 cm peak-to-peak motion (A0), a 1 cm internal margin was used for CTV expansion to ITV (ITV1 cm), and similarly, a 2 cm internal margin was used as the target volume for a 2.0 cm peak-to-peak motion: z() = A0 cos4 ( ). (2) Irregular motion proles used in the study were created similarly by generating 100 respiratory cycles which follow the functional form in equation (2). However, each respiratory

Irregular respiratory motion

851

Figure 4. Plot of the functional form of respiration used in the simulations. The peak-to-peak amplitude of the motion, A0, is used as the free parameter and adjusted accordingly as described in the text.

cycle within a session was varied by changing its peak-to-peak amplitude (A0) sampled from a Gaussian distribution with a preselected mean (A) and standard deviation ( A). The examples of simulated irregular respiration cycles and the corresponding distribution of peak-to-peak amplitudes are shown in gure 5. Irregular respiratory motion simulations are further divided into two categories. Characteristic or random irregular motion where the irregularities in the respiratory motion during treatment are distributed randomly about the motion amplitude attributed to the CTV during the 4DCT imaging. Uncharacteristic or systematically irregular motion where there is a systematic shift between CTV motion amplitude determined at the time of 4DCT imaging and the actual respiratory behavior of the patient during treatment. In characteristic motion, the mean of the Gaussian distribution for peak-to-peak amplitudes is chosen to be equal to the internal margin. For example, a set of respiratory cycles distributed about a mean of 1.0 cm (e.g. gure 5, upper panel) is characteristic for the simulations of CTV motion using the plan generated with a 1.0 cm internal margin (ITV1 cm). In uncharacteristic motion simulations, however, the internal margin between the CTV and ITV is smaller than the mean of the respiratory cycle amplitudes by 1 standard deviation. An example to such respiratory motion would be a set of respiratory cycles distributed according to a mean of 1.5 cm and a standard deviation of 0.5 cm (e.g. gure 5, lower panel) used for the simulations with only a 1.0 cm internal margin (ITV1 cm). A characteristic irregular motion would, therefore, be the best-case scenario where the irregularities are essentially random in nature and CTV motion as observed in the simulation (e.g. with 4DCT) corresponds to the average target motion. In contrast, the uncharacteristic

852

Y D Mutaf et al

Figure 5. A time series plot of the simulated irregular respiration used in the study for the mean peak-to-peak amplitude values (A) of 1.0 cm (upper panel) and 1.5 cm (lower panel). A histogram of the cycle peak-to-peak amplitudes is plotted next to the corresponding plot of the respiration data. Although the continuous form of the full respiration data is shown (solid line), the simulations are performed for respiration amplitudes sampled at phase intervals of 10% (lled circles). The sampled motion is also digitized by 1.25 mm equal to the slice thickness of the CT images.

irregular motion reects a more extreme case where in addition to random variations in respiration, the determined CTV motion and internal margin are compromised by a systematic difference. This systematic error occurs when the amount of motion determined at the planning stage is consistently less than the average target motion observed during patients treatment. Characteristic and uncharacteristic irregular motion simulation parameters are listed below. Simulations with ITV1 cm A = 1.00 cm, A = 0.25 cm or 25% (characteristic) A = 1.00 cm, A = 0.50 cm or 50% (characteristic) A = 1.25 cm, A = 0.25 cm (uncharacteristic) A = 1.50 cm, A = 0.50 cm (uncharacteristic). Simulations with ITV2 cm A = 2.00 cm, A = 0.50 cm or 25% (characteristic) A = 2.00 cm, A = 1.00 cm or 50% (characteristic) A = 2.50 cm, A = 0.50 cm (uncharacteristic) A = 3.00 cm, A = 1.00 cm (uncharacteristic). These a priori choices of the mean and the standard value of the respiration amplitude distribution are intended to provide several threshold levels for their inuence on target dosimetry rather than exact representations of what is observed with actual patients. In this respect, 25% variations as used in this study stand as a close approximation to what we observed within our patient population as well as what is reported by other investigators (Britton et al 2007). However, amplitude variations as large as 50% are not typically reported in the literature and therefore constitute a limiting case.

Irregular respiratory motion

853

Figure 6. Respiratory motion trace of a patient (abdomen motion). The average peak inhale and exhale amplitudes are emphasized with blue and red horizontal lines, respectively.

In addition to the functional forms of irregular respiratory motion as described above, patients actual respiratory waveforms were also used for target motion simulations; however, these data were available only for a subset of patients, 13 out of 23 patients studied here. The respiratory signal represented the vertical motion of the abdominal surface recorded with the aid of reective markers placed on patients abdomen and an infrared camera as available in a commercial respiratory management system (RPM, Varian Inc.). In clinical 4DCT procedures, these patients were subjected to audio-coaching feedback methods for improving the reproducibility of their respiratory pattern (Kini et al 2003, George et al 2006, Haasbeek et al 2008). The abdominal motion is used as surrogate for the actual tumor motion (Rietzel et al 2005, Mutaf and Brinkmann 2008a) and has been shown to have a high correlation with the actual tumor motion in lung cancer patients (Ozhasoglu and Murphy 2002, Starkschall et al 2004, Gierga et al 2005). However, this correlation is only relevant for the relative amounts of surrogate and target motions and therefore the surrogate motion is scaled in the simulations such that the average respiratory cycle from the entire surrogate waveform was normalized to selected peak-to-peak amplitudes. As described earlier, this normalization was chosen as 1.0 cm for the average peak-to-peak motion of the surrogate data in simulations for plans with ITV1 cm and 2.0 cm for simulations with ITV2 cm. An example of patient respiration waveform is demonstrated in gure 6. The average respiration cycle data are used for regular motion simulation; however, the full respiration waveform is used for irregular motion simulations. In contrast to the synthesized respiratory motion data used for characteristic and uncharacteristic irregular motion simulations, the actual respiratory trace of the patients provides a more diverse spectrum of irregularities in breathing. While synthesized respiratory simulations only represent variations in inhale amplitude, actual motion data also include such irregular motion as variations in end-of-exhale (a.k.a. baseline-drift) as well as the presence of outlier respiratory cycles. However, due to the use of single session data, the simulations of irregular motion with patient motion traces only represent intra-fractional variations excluding possible inter-fractional changes in breathing. 2.4. Plan evaluation metrics Several metrics are utilized to assess the effect of irregular respiratory motion on dosimetric coverage for CTV. Respiratory motion mostly affects the cells at the margins of the target causing them to move outside the boundaries of the irradiation eld and receive less than

854

Y D Mutaf et al

planned dose. Therefore, the target minimum dose (Dmin) is expected to be the most sensitive metric for dosimetric evaluation of structures under motion. In addition, global dosimetric evaluators such as the equivalent uniform dose (EUD) (Niemierko 1997, Choi and Deasy 2002) and tumor control probability (TCP) (Webb and Nahum 1993, Okunieff et al 1995) are used for evaluations. The EUD is dened such that if a target is irradiated uniformly with this dose, it will result in the same clonogenic cell killing as the actual inhomogeneous dose distribution across the target volume. Assuming a homogeneous density of tumor clonogens within the target volume and ignoring cell repair and repopulation effects, Niemierkos (1997) EUD formalism was adopted for this study as shown in equation (3): (SF2 )Di /Dref . (3) ln(SF2 ) In equation (3), the reference dose, Dref, is dened as the reference dose per fraction and usually taken as 2 Gy, whereas SF2, the surviving fraction of clonogens at 2 Gy, is set to 0.5. Using these parameters, the average EUD values were 61.1 Gy (range 59.7 to 63.1 Gy) and 60.4 Gy (range 59.5 to 61.2 Gy) for simple and complex plans, respectively (note that the all prescription doses have been normalized to 60 Gy in all patient plans for the purpose of this study). TCP is also calculated to assist the dosimetric evaluations for the target and assess any dosimetric change due to irregular respiratory motion in terms of its relative role in undermining local control, if any. TCP is dened as the Poisson probability for achieving complete tumor cell lethality (zero survival probability). Analysis of dose-response data from populationbased studies results in several empirical formulations for TCP and we implemented the logit form as described in Okunieff et al (1995) and shown in equation (4): EUD = Dref TCP = e 450 DD50 D50 1 + e 450 DD50 D50 . (4) ln
1 N N i=1

Tumor control probability has a sigmoid shape which is modeled by the two free parameters D50 and 50, dose to achieve 50% control probability and the slope of the sigmoid curve at D50, respectively. Following the multi-institutional analysis of the dose response data for macroscopic lung tumors of Okunieff et al we adopted a D50 value of 51.2 Gy and 50 of 0.83. For this set of parameters, the average TCP values were 0.66 (range 0.640.69) and 0.65 (range 0.630.66) for simple and complex plans, respectively. It is not the intent of the current study to evaluate the validity of these treatment plan metrics or their selected parameters. These metrics are solely used to facilitate comparisons between target dosimetry under irregular respiratory motion as compared to its regular motion. Therefore, these metrics are only interpreted with respect to their relative values within corresponding comparisons. Furthermore, we dene a gure of relativity for each metric discussed above that is equal to the ratio of the same metric for the target under irregular motion to that under regular motion. This is formulated in equation (5). Following this approach, we dene rel-Dmin, rel-EUD and rel-TCP metrics for the targets and motions studied here: rel-M = Mirreg /Mreg . 3. Results The result of characteristic irregular motion simulations with 1.0 and 2.0 cm amplitudes are displayed in gure 7 for all 23 patients. The changes in the target minimum dose, represented by rel-Dmin, are also shown for simple and complex plans. (5)

Irregular respiratory motion

855

Figure 7. Relative target minimum dose (rel-Dmin) results of characteristic irregular motion simulations for all patients; mean motion amplitudes of 1 cm (upper plot) and 2 cm (lower panel) are shown.

In characteristic motion simulations with a 1 cm average amplitude and 25% standard deviation, the maximum change in the minimum dose to CTV was 0.1% for both simple and complex plans. However no change in EUD or TCP was observed. A change in target dose was slightly larger when the variations in respiratory motion increased to 50% standard deviation with a maximum reduction of 0.6% in the minimum target dose for simple and 0.8% for complex plans. For this motion, maximum changes in both EUD and TCP were 0.2% in both simple and complex plans among all patients. When the motion amount was increased to 2 cm in amplitude, changes in all dosimetry metrics for CTV were larger for both 25% and 50% variations. The maximum change in the minimum dose to a target was 3.7% in the case of 50% variations. For 50% variations, the maximum EUD reduction was 2.1% and the TCP reduction was 3.0%. No substantial difference between simple and complex plan dosimetry was observed. Target dosimetry for uncharacteristic motion is also presented in comparison with the characteristic motion in gure 8 shown only for complex geometry plans. The effect of uncharacteristic motion on simple plans again was not substantially different than the complex plans. For plans with 1.0 and 2.0 cm internal margins in all patients, CTV dose was compromised more drastically in uncharacteristic irregular motion simulations compared to characteristic motion. The largest change in target dosimetry was observed for plan with a 2.0 cm margin and associated uncharacteristic irregular motion ( = 3.0 cm,

856

Y D Mutaf et al

Figure 8. Comparison of relative target minimum dose (rel-Dmin) in irregular motion simulations for characteristic and uncharacteristic motion traces; simulations with 1 cm (upper panel) and 2 cm (lower panel) internal margins are shown.

= 1.0 cm) such that the minimum target dose was reduced by 13.6% as compared to the regular motion. Simulations of respiratory motion using the actual abdominal (surrogate) motion of patients were also repeated for both 1.0 and 2.0 cm average motion amplitudes. The maximum change in the target minimum dose was 0.1% in both cases using complex plans. The maximum reduction in EUD was 0.3% and in TCP was 0.5% observed only with a 2 cm motion. Simple plans yielded similar results. The ranges of target dosimetry metrics among all the patients for all simulated motions are summarized in table 2. 4. Discussion In this study, an evaluation of the dosimetric effects of irregular motion was presented for 23 lung cancer patients. Several types of irregular motion simulations are performed to assess target dosimetry under various irregular motion conditions. In the process, the interplay between the plan topology as demonstrated by simple and complex three-dimensional dosimetry and the presence of irregular respiratory motion was also investigated. Characteristic irregular motion type representative of random variations in the determined (i.e. planned) target motion caused non-substantive dosimetric effects. In most simulated cases,

Irregular respiratory motion

857

Table 2. Range of target dosimetry evaluation metrics for the patient population investigated in this study. Relative metrics as tabulated here are dened in equation (5).

Characteristic motion (IM = ) = 1.0 cm, = 1.0 cm, = 2.0 cm, = 2.0 cm, = 0.25 cm (IM = 1.0 cm) = 0.50 cm (IM = 1.0 cm) = 0.5 cm (IM = 2.0 cm) = 1.0 cm (IM = 2.0 cm)

rel-Dmin 99.9100.0% 99.2100.0% 99.5100.0% 96.299.6% 99.0100.0% 91.699.3% 95.8100.0% 86.493.6% 99.9100.0% 99.0100.0%

rel-EUD 100.0% 99.8100.0% 99.8100.0% 97.999.8% 99.8100.0% 98.299.9% 98.799.9% 92.799.0% 99.9100.0% 99.7100.0%

rel-TCP 100.0% 99.8100.0% 99.7100.0% 97.099.7% 99.8100.0% 98.199.8% 98.499.9% 90.899.0% 99.9100.0% 99.5 100.0%

Un-characteristic motion (IM = ) = 1.25 cm, = 0.25 cm (IM = 1.0 cm) = 1.50 cm, = 0.50 cm (IM = 1.0 cm) = 2.5 cm, = 0.5 cm (IM = 2.0 cm) = 3.0 cm, = 1.0 cm (IM = 2.0 cm) Patient respiratory motion Normalized to 1 cm (IM = 1.0 cm) Normalized to 2 cm (IM = 2.0 cm)

the change in the target minimum dose was clinically insignicant (less than 2%) to require intervention or other compensation techniques. In the extreme scenario where the simulated irregular motion had a mean of 2 cm and showed 1 cm amplitude variations (at 1 standard deviation), the minimum target dose was dropped by 2.5% 0.9%. Although minimal, these observations conrm the expectation that the dosimetric effect of characteristic irregularities increases as the average amount of motion and its relative variations increases. However, there is no assurance that the target motion as determined during patient simulation (e.g. 4DCT) would be reective of the average respiratory motion for the entire course of treatment. For normally distributed motion data, there is a 50% probability that the internal margin amount determined at simulation will be smaller than the average amplitude of the motion. Dosimetric consequences for such scenarios are investigated in this study via the simulations of uncharacteristic irregular motion where the chosen internal margin for the target is smaller than the mean of the entire peak amplitude distribution by 1 standard deviation. Uncharacteristic irregular motion resulted in clinically signicant dosimetric consequences with minimum CTV doses reduced about 8.4% for a 1.0 cm internal margin (motion = 1.5 cm, = 0.5 cm) and up to 13.6% for a 2.0 cm internal margin (motion = 3.0 cm, = 1.0 cm). This observation signies that such systematic discrepancies between the determined target motion and the entire respiratory motion have more impact than the random variations in respiration. In fractionated radiotherapy, an example of this discrepancy could appear as inter-fractional variations of the respiratory motion between the simulation session where target motion is determined and subsequent treatment sessions. Results from the simulation of actual patient respiratory waveforms indicated that the respiration motion irregularities demonstrated by these patients did not cause detrimental dosimetric consequences. This parallels our observations with characteristic irregular motion. The similarity between the actual patient respiration motion and generated characteristic irregular motion simulations could be attributed to the fact that the patient respiratory data were collected with the help of audio-coaching techniques. Since coaching compels the patient to reproduce a respiratory pattern, the breathing irregularities are consequentially restricted to random uctuations. Therefore, it can be argued that the coaching techniques help avoid

858

Y D Mutaf et al

the occurrence of uncharacteristic respiratory motion which was observed to result in largest degradation for target dosimetry. 5. Conclusion Determination of respiratory motion is a necessary rst step for the management of respiratory motion in radiotherapy. The process also requires that determined respiratory motion is valid throughout the full course of patients treatment. However, this validity is compromised due to the varying degrees of breathing irregularities. In this study, we assessed the clinical implications and dosimetric consequences of these variations and categorized different types of breathing irregularities. Characteristic motion irregularities are not observed to have a considerable clinical effect on target dosimetry. Similarly, simulations of target motion using actual patient respiration data have not produced a clinically substantial dosimetric effect. However, uncharacteristic irregularities are shown to have critical dosimetric consequences for the target coverage. Coaching techniques are suggested as potential ways to avoid such inconsistencies and maintain a common baseline for patients respiration throughout the course of radiotherapy. Additionally, there was no substantial difference between the simple and complex treatment plan geometries incorporating different levels of target dose homogeneity and conformity. References
Britton K R, Starkschall G, Tucker S L, Pan T, Nelson C, Chang J Y, Cox J D, Mohan R and Komaki R 2007 Assessment of gross tumor volume regression and motion changes during radiotherapy for non-small-cell lung cancer as measured by four-dimensional computed tomography Int. J. Radiat. Oncol. Biol. Phys. 68 103646 Chang J et al 2007 Observation of interfractional variations in lung tumor position using respiratory gated and ungated megavoltage cone-beam computed tomography Int. J. Radiat. Oncol. Biol. Phys. 67 154858 Choi B and Deasy J O 2002 The generalized equivalent uniform dose function as a basis for intensity-modulated treatment planning Phys. Med. Biol. 47 357989 George R, Chung T D, Vedam S S, Ramakrishnan V, Mohan R, Weiss E and Keall P J 2006 Audio-visual biofeedback for respiratory-gated radiotherapy: impact of audio instruction and audio-visual biofeedback on respiratorygated radiotherapy Int. J. Radiat. Oncol. Biol. Phys. 65 92433 Gierga D P, Brewer J, Sharp G C, Betke M, Willett C G and Chen G T 2005 The correlation between internal and external markers for abdominal tumors: implications for respiratory gating Int. J. Radiat. Oncol. Biol. Phys. 61 15518 Graham M V, Purdy J A, Emami B, Harms W, Bosch W, Lockett M A and Perez C A 1999 Clinical dose-volume histogram analysis for pneumonitis after 3D treatment for non-small cell lung cancer (NSCLC) Int. J. Radiat. Oncol. Biol. Phys. 45 3239 Haasbeek C J, Spoelstra F O, Lagerwaard F J, van Sornsen de Koste J R, Cuijpers J P, Slotman B J and Senan S 2008 Impact of audio-coaching on the position of lung tumors Int. J. Radiat. Oncol. Biol. Phys. 71 111823 International Commission on Radiation Units and Measurements 1993 Prescribing, recording, and reporting photon beam therapy ICRU Report 50 (Bethesda, MD: ICRU) International Commission on Radiation Units and Measurements 1999 Prescribing, recording, and reporting photon beam therapy ICRU Report 62 (Bethesda, MD: ICRU) Juhler Nottrup T, Korreman S S, Pedersen A N, Aarup L R, Nystrom H, Olsen M and Specht L 2007 Intra- and interfraction breathing variations during curative radiotherapy for lung cancer Radiother. Oncol. 84 408 Kini V R, Vedam S S, Keall P J, Patil S, Chen C and Mohan R 2003 Patient training in respiratory-gated radiotherapy Med. Dosim. 28 711 Lujan A E, Balter J M and Ten Haken R K 2003 A method for incorporating organ motion due to breathing into 3D dose calculations in the liver: sensitivity to variations in motion Med. Phys. 30 26439 Murphy M J, Martin D, Whyte R, Hai J, Ozhasoglu C and Le Q T 2002 The effectiveness of breath-holding to stabilize lung and pancreas tumors during radiosurgery Int. J. Radiat. Oncol. Biol. Phys. 53 47582 Mutaf Y D and Brinkmann D H 2008a An investigation of temporal resolution parameters in cine-mode fourdimensional computed tomography acquisition J. Appl. Clin. Med. Phys. 9 2819

Irregular respiratory motion

859

Mutaf Y D and Brinkmann D H 2008b Optimization of internal margin to account for dosimetric effects of respiratory motion Int. J. Radiat. Oncol. Biol. Phys. 70 156170 Niemierko A 1997 Reporting and analyzing dose distributions: a concept of equivalent uniform dose Med. Phys. 24 10310 Oetzel D, Schraube P, Hensley F, Sroka-Perez G, Menke M and Flentje M 1995 Estimation of pneumonitis risk in three-dimensional treatment planning using dose-volume histogram analysis Int. J. Radiat. Oncol. Biol. Phys. 33 45560 Okunieff P, Morgan D, Niemierko A and Suit H D 1995 Radiation dose-response of human tumors Int. J. Radiat. Oncol. Biol. Phys. 32 122737 Ozhasoglu C and Murphy M J 2002 Issues in respiratory motion compensation during external-beam radiotherapy Int. J. Radiat. Oncol. Biol. Phys. 52 138999 Ramsey C R, Scaperoth D, Arwood D and Oliver A L 1999 Clinical efcacy of respiratory gated conformal radiation therapy Med. Dosim. 24 1159 Rietzel E, Pan T and Chen G T 2005 Four-dimensional computed tomography: image formation and clinical protocol Med. Phys. 32 87489 Seppenwoolde Y, Shirato H, Kitamura K, Shimizu S, van Herk M, Lebesque J V and Miyasaka K 2002 Precise and real-time measurement of 3D tumor motion in lung due to breathing and heartbeat, measured during radiotherapy Int. J. Radiat. Oncol. Biol. Phys. 53 82234 Shaw E, Scott C, Souhami L, Dinapoli R, Kline R, Loefer J and Farnan N 2000 Single dose radiosurgical treatment of recurrent previously irradiated primary brain tumors and brain metastases: nal report of RTOG protocol 9005 Int. J. Radiat. Oncol. Biol. Phys. 47 2918 Starkschall G, Forster K M, Kitamura K, Cardenas A, Tucker S L and Stevens C W 2004 Correlation of gross tumor volume excursion with potential benets of respiratory gating Int. J. Radiat. Oncol. Biol. Phys. 60 12917 van Herk M 2004 Errors and margins in radiotherapy Semin. Radiat. Oncol. 14 5264 van Herk M, Witte M, van der Geer J, Schneider C and Lebesque J V 2003 Biologic and physical fractionation effects of random geometric errors Int. J. Radiat. Oncol. Biol. Phys. 57 146071 Wagman R, Yorke E, Ford E, Giraud P, Mageras G, Minsky B and Rosenzweig K 2003 Respiratory gating for liver tumors: use in dose escalation Int. J. Radiat. Oncol. Biol. Phys. 55 65968 Webb S and Nahum A E 1993 A model for calculating tumour control probability in radiotherapy including the effects of inhomogeneous distributions of dose and clonogenic cell density Phys. Med. Biol. 38 65366