Development and application of a real‐time monitoring and feedback ...

Development and application of a real-time monitoring and feedback system for deep inspiration breath hold based on external marker tracking Markus Stock,a兲 Kristina Kontrisova, Karin Dieckmann, Joachim Bogner, Richard Poetter, and Dietmar Georg Department of Radiotherapy and Radiobiology, AKH Vienna, Medical University Vienna, Vienna, Austria

共Received 6 February 2006; revised 12 June 2006; accepted for publication 14 June 2006; published 24 July 2006兲 Respiration can cause tumor movements in thoracic regions of up to 3 cm. To minimize motion effects several approaches, such as gating and deep inspiration breath hold 共DIBH兲, are still under development. The goal of our study was to develop and evaluate a noninvasive system for gated DIBH 共GDIBH兲 based on external markers. DIBH monitoring was based on an infrared tracking system and an in-house-developed software. The in-house software provided the breathing curve in real time and was used as on-line information for a prototype of a feedback device. Reproducibility and stability of the breath holds were evaluated without and with feedback. Thirty-five patients undergoing stereotactic body radiotherapy 共SBRT兲 performed DIBH maneuvers after each treatment. For 16 patients dynamic imaging sequences on a multislice CT were used to determine the correlation between tumor and external markers. The relative reproducibility of DIBH maneuvers was improved with the feedback device 共74.5% ± 17.1% without versus 93.0% ± 4.4% with feedback兲. The correlation between tumor and marker was good 共Pearson correlation coefficient 0.83± 0.17兲. The regression slopes showed great intersubject variability but on average the internal margin in a DIBH treatment situation could be theoretically reduced by 3 mm with the feedback device. DIBH monitoring could be realized in a noninvasive manner through external marker tracking. We conclude that reduction of internal margins can be achieved with a feedback system but should be performed with great care due to the individual behavior of target motion. © 2006 American Association of Physicists in Medicine. 关DOI: 10.1118/1.2219775兴 Key words: real-time monitoring, feedback system, DIBH, gated radiotherapy, external markers I. INTRODUCTION The aim of all advanced treatment techniques is to reduce normal tissue toxicity and to escalate the dose to the tumor at the same time. Especially, conformal and intensity modulated radiotherapy 共CRT, IMRT兲 addressed this issue in the past decade. Additional improvements, in terms of reduced toxicity, can be achieved by decreasing margins because of benefits in normal tissue sparing, which can be reinvested in dose escalation.1–3 The results of conformal radiotherapy for targets in the lung are still not satisfying4,5 and therefore clinical benefits of margin reductions can be expected. Certainly, one has to pay special attention to the increasing risk of missing the target and additional 共technological兲 effort is necessary to minimize internal margins 共IM兲.6 Several solutions have already been presented to account for respiratory motion. The first class of approaches is the so-called gating technique. Gating signals, which correspond to the position of the tumor, are used to trigger the linear accelerator and also computed tomography 共CT兲, magnetic resonance imaging 共MR兲, and positron emission tomography 共PET兲 scanners in order to minimize motion artifacts. As signal sources, external7–11 or internal markers12–14 as well as other devices like spirometer,15,16 strain gauge,17 laser sensor system,18 and temperature sensor17 are in use. The gating system is mostly used for applying burst of radiation during the normal free2868

Med. Phys. 33 „8…, August 2006

breathing phases. One can either define the beam on time or a beam on window for certain amplitudes of the breathing cycle. The second class of approaches is the breath hold technique, where the patient holds his breath at certain phases of the breathing cycle with active or passive control devices1–3,7,19–26 or even without a control device.27,28 In most cases control devices are similar to gating devices but are not used to trigger the linear accelerator. There are two different existing philosophies, mainly breath hold at expiration8,27 or deep expiration 共DEBH兲 共Ref. 27兲 in contrast to inspiration27 or deep inspiration 共DIBH兲.1,2,24–27 Several groups cited above performed measurements concerning reproducibility of the breath hold phases, either for a patient self-breath holding,28 by using a spirometer signal as feedback, or by blocking the patient’s breath with an active breathing control device 共ABC兲. The third class integrates approaches using a frame or frameless vacuum pillow for better repositioning and in addition abdominal pressure for restricted respiration.29–32 These techniques are mainly used for hypofractionated stereotactic body radiotherapy 共SBRT兲. In addition to the body frame an abdominal compression is used to minimize the respiration movement and on the other hand to assist the patient to perform shallow breathing. Promising clinical results were already obtained concerning local control rates.33,34

0094-2405/2006/33„8…/2868/10/$23.00

© 2006 Am. Assoc. Phys. Med.

2868

2869

Stock et al.: Real-time monitoring of external marker for gated DIBH

2869

FIG. 1. 共a兲 Feedback device at a typical training situation of a volunteer on the treatment couch with indication of the marker positions. The device consists of a 7 in. thin-film transistor 共TFT兲 flat-panel display which was rigidly attached to a holder. 共b兲 Screenshot of the feedback during a DIBH.

Our aim was to develop and evaluate a gated deep inspiration breath hold 共GDIBH兲 procedure based on external markers. We intended for good repositioning results of patient and tumor in a noninvasive way using a feedback system, where the patient can reproduce DIBHs by controlling the movement of the thorax. Before implementing the technique into clinical routine, we tried to find out how well external markers correlate with the tumor movement, quantified the effect of feedback device, and determined the achievable treatment margins for SBRT.

II. MATERIALS AND METHODS A. Patient characteristics

The current study was performed with 35 patients 共12 female, 23 male兲. The patients suffered either from lung tumors 共2 patients兲 or from lung 共24 patients兲 or liver 共9 patients兲 metastases. The mean age was 61.5 years 共range 21 to 81兲. All patients were able to understand breathing instructions, were able to perform a breath hold for at least 10 seconds, and were willing to participate in the study. Only patients with a Karnofsky performance status 艌90 were considered.

B. Patient immobilization

All patients underwent SBRT. SBRT at the Medical University of Vienna is based on a stereotactic body frame 共SBF, Elekta, Crawley, UK兲 originally developed by Lax and Blomgren.29 A detailed description of the system and its use is given in a work of Wulf et al.31 The SBF has been modified in the department’s mechanical workshop to integrate an arm rest, which is attached to a modified base plate of the SBF. The design is comparable to a T-bar as investigated by Halperin et al.35 Medical Physics, Vol. 33, No. 8, August 2006

C. DIBH monitoring and feedback system

The DIBH monitoring and feedback system was developed based on the ExacTrac® system, Beta-Version 3.5 共BrainLAB, Heimstetten, Germany兲.36 In the current infrared 共IR兲 tracking system software 共server application兲 a dynamic data exchange 共DDE兲 server was implemented in order to be able to extract positional information in all three directions of each individual marker in real time. A feedback software 共client application兲 was developed in-house to visualize the anterior-posterior 共AP兲 and the cranial-caudal 共CC兲 marker movements. All marker positions are transferred, updated, and stored each 200 ms in the feedback software. Mainly four markers were used for DIBH monitoring and were placed on the costal arch, xiphoid process, near the tumor position on the thorax, and on the abdomen 关see Fig. 1共a兲兴. All marker positions were indicated on the patient using a skin marker pen. During the first training session a representative marker signal 共mostly AP or CC of xiphoid process marker兲 for feedback and monitoring purposes was selected. This marker is addressed as the “tracking marker” in the following. Figure 1 shows the feedback device with the marker configuration described above and a screenshot of the feedback signal during a DIBH. The patients were instructed to hold their breath as long as indicated on the screen 关see the lines in Fig. 1共b兲兴. After the elapsed duration the lines disappear and the patient can perform shallow breathing again. In addition, the tolerance level is visualized 关see Fig. 1共b兲兴. The tolerance region around the DIBH amplitude and the breath hold duration can be defined individually by the operator with the graphical user interface 共initial setting ±1 mm兲. Another software feature enables the user to reload the breathing curve after the session from the hard disk and to analyze it in more detail. D. DIBH stability and reproducibility

In the very first session, before undergoing CT, patients were asked to perform shallow breathing 共SB兲, and then they

2870

Stock et al.: Real-time monitoring of external marker for gated DIBH

2870

FIG. 2. Signals with different signal qualities Q. Q was defined as the ratio between aref and peak-to-peak amplitude during SB. Signal 共c兲 has the best signal quality.

were instructed to perform a DIBH. This so-called setup procedure provided information of the mean marker position during SB and DIBH. The positional difference of the tracking marker between mean SB and deep inspiration in one direction 共AP or CC兲 was used as reference amplitude for all other sessions 共aref兲. To quantify DIBH reproducibility and stability, mean position values and standard deviations of the tracking marker, and those of all other markers, were calculated. The mean value of the standard deviation was used to quantify intrabreath hold stability during DIBH. The intrabreath hold stability is a measure for the knowing and unknowing muscle relaxation during the breath hold. The average of all absolute differences between mean DIBH positions with respect to the reference amplitude was defined as DIBH deviation d, n

兺 abs共ai − aref兲

d=

i=1

n

,

where d is deviation, ai is mean value of the ith DIBH amplitude, aref is reference amplitude of setup procedure, and n is number of DIBHs. Therefore, the reproducibility r can be defined as r = aref − d. For each marker signal both reproducibility and stability were normalized to aref. These relative reproducibility and Medical Physics, Vol. 33, No. 8, August 2006

FIG. 3. Indication of measurement procedure for the movement of the marker in AP direction for patient 1 in all dynamic scans 共3 s between scans兲. The distance from the ground plate of the SBF to the marker is indicated below each axial slice. The position of the marker in CC direction can be determined by the slice number. See also Fig. 4.

stability values 共of each marker signal兲 were averaged over all patients to determine the impact of the feedback system. E. Signal quality for feedback

A signal quality factor 共Q兲 was introduced, which was defined as the ratio between aref and peak-to-peak amplitude during SB. To avoid errors or misinterpretation during GDIBH, the SB inspiration peak should be well distinguishable from the DIBH plateau of the tracking signal. Thus, Q is an indicator for the usefulness of a signal for DIBH monitoring and as feedback during DIBH maneuvers. In general, the better the signal quality 共high Q value兲 the larger is the difference between SB and deep inspiration 共see Fig. 2兲. Relative reproducibility and stability were evaluated with respect to signal quality. The signal quality in turn was evaluated as a function of marker position and direction of movement. F. Marker-tumor correlation

Sixteen out of the 35 patients were included in a multislice CT investigation which was conducted to determine the

2871

Stock et al.: Real-time monitoring of external marker for gated DIBH

2871

TABLE I. Relative reproducibility and stability of tracking marker and other markers due to signal quality, with or without feedback. In the first and third column the reproducibility was evaluated independent of the signal quality index Q. In the second and fourth column only signals with a Q better than 2 were evaluated. FB is feedback, Q is signal quality index.

Mean rel. reproducibility + / −1 SD Mean rel. stability + / −1 SD

Tracking marker

Tracking marker 共Q ⬎ 2兲

Other markers

Other markers 共Q ⬎ 2兲

AP

AP

AP

AP

without FB

74.5+ / −17.1%

74.5+ / −17.5%

69.5+ / −33.5%

76.0+ / −15.2%

with FB

93.0+ / −4.4%

93.2+ / −4.3%

72.2+ / −26.6%

77.5+ / −14.8%

without FB

5.0+ / −3.0%

4.9+ / −3.0%

5.3+ / −3.8%

4.6+ / −3.0%

with FB

4.2+ / −2.7%

4.2+ / −2.7%

5.3+ / −4.6%

4.3+ / −2.9%

correlation between one external marker, referred to as “tumor marker,” and the target. This investigation was performed using a multislice CT Somatom Sensation 16 共Siemens, Erlangen, Germany兲. Patients were positioned in the SBF similar as for treatment. The movements during several respiratory cycles in SB and during DIBH and DEBH were investigated in the dynamic 3D scan mode 共fixed table position兲 every 3 s. For the scan the slice collimator was set to 3 ⫻ 1.5, the table feed was 0 mm/ s, and the slice width was 3 mm with a reconstruction increment of 3 mm. The gantry rotation time was always 0.5 s. Using this configuration, it was possible to investigate the thorax/tumor marker and the tumor in a scan volume of 24 mm length. In total, mostly five scans were acquired in SB, four scans in DIBH and four scans in DEBH, with the remaining scans being acquired between these breathing phases. All scans were transferred to the BRAINSCAN® planning software 共BrainLAB, Heimstetten, Germany兲. The first scan was used as a reference. All other scans were manually registered with respect to the reference data set with the image fusion tool of the planning software using the tumor shape as a template for the best fit. The CT data shift in all three directions was taken as the three-dimensional tumor movement with respect to the reference scan. Based on the dynamic scan data sets, the movement of the tumor marker was measured using the base plate of the SBF as reference 共see Fig. 3兲.

For further analysis tumor 共T兲 movements in AP or CC direction were quantitatively correlated with marker 共M兲 movements in AP or CC direction using the Pearsoncorrelation formula, Rpear =

冉 冊冉

n 1 Ti − ¯T 兺 n − 1 i=1 sT

冊

¯ Mi − M , sM

where Ti is tumor position at i’th dynamic scan, M i is marker position at i’th dynamic scan, ST is standard deviation of tumor position, S M is standard deviation of marker position, ¯T is mean tumor position, and M ¯ is mean marker position. Correlation coefficients were calculated for the following combinations: T-AP vs M-AP, T-AP vs M-CC, T-CC vs M -AP, and T-CC vs M-CC. A linear regression was fitted through data points of tumor and marker positions and the slopes were analyzed for each patient. III. RESULTS A. DIBH stability and reproducibility

In total, 157 breathing sessions were performed, with 313 DIBH maneuvers without feedback and 524 with feedback system. Ninety-four DIBH maneuvers with feedback were not evaluated because patients were not able to stay within the default tolerance interval of ±1 mm during the breath

TABLE II. Amplitude of marker movement during shallow breathing and amplitude of DIBH in mm, as well as signal quality depending on the location of the marker. SB is shallow breathing, Q is signal quality index, SD is standard deviation. Thorax

Mean amplitude of SB 共+ / −SD兲 关mm兴 Mean amplitude of DIBH 共+ / −SD兲 关mm兴 Mean Q 共+ / −SD兲

Xiphoid process

Costal arch

Abdomen

CC

AP

CC

AP

CC

AP

CC

AP

1.8 共+ / −0.9兲

1.6 共+ / −0.8兲

1.9 共+ / −0.9兲

1.6 共+ / −0.8兲

2.3 共+ / −1.3兲

1.8 共+ / −1.0兲

1.7 共+ / −0.9兲

5.6 共+ / −4.0兲

5.8 共+ / −3.6兲

6.9 共+ / −3.3兲

6.5 共+ / −4.2兲

6.9 共+ / −3.3兲

6.1 共+ / −3.7兲

5.8 共+ / −2.7兲

3.8 共+ / −3.3兲

11.2 共+ / −7.9兲

4.0 共+ / −3.3兲

5.8 共+ / −4.6兲

4.2 共+ / −3.3兲

5.5 共+ / −3.5兲

3.4 共+ / −2.5兲

4.6 共+ / −3.9兲

2.9 共+ / −1.9兲

2.4 共+ / −1.4兲

Medical Physics, Vol. 33, No. 8, August 2006

2872

Stock et al.: Real-time monitoring of external marker for gated DIBH

hold. This was the first or second breath hold with feedback in most cases. After the first two maneuvers nearly every patient stayed in the tolerance interval. This reflects the learning curve with the system. On average, each patient performed 9 DIBH maneuvers without and 12 with feedback system in total of 4–5 sessions 共one session before the planning CT and the remaining immediately after each treatment fraction兲. The relative DIBH reproducibility of the tracking marker using the feedback was 93.0% ± 4.4% compared to 74.5% ± 17.1% without feedback 共see Table I兲. The reproducibility of other markers 共excluding the tracking marker兲 in AP direction is shown as well in Table I. If only signals with a quality better than 2 are considered, the mean value can be increased and the standard deviation can be reduced. In other words the signal quality 共by choosing only useful signals兲 has a direct impact on the reproducibility. The tracking marker almost always had a better signal quality than 2. The difference between the intrabreath hold stability with and without feedback can be neglected. The relative stability of the feedback signal was 5.0± 3.0% without feedback and 4.2± 2.7% with feedback, respectively 共see Table I兲. In this study patients were asked to hold their breath for only 10 to 12 s 共mean value 11 s兲 and the signal stability may increase for longer DIBH maneuver. When investigating the stability of all breath holds 共with and without feedback兲 as a function of marker position, a small difference can be observed between the stability in CC direction of the abdomen marker and all other marker locations. The relative intrabreath hold stability for the abdomen marker was 9.5± 10.2% in CC but 3.7± 2.7% in AP direction. The relative stability for costal arch, xiphoid process, and thoracic marker was the same in AP direction but 50% smaller in CC direction. Mean marker amplitudes during SB and DIBH determined from the setup procedure are listed in Table II. The largest amplitudes during shallow breathing and deep inspiration were observed in AP direction for markers placed on the abdominal region.

2872

FIG. 4. Tumor and tumor marker movement during the multislice session measured with dynamic scan mode for patient 1. M-AP is marker movement in AP direction, M-CC is marker movement in CC direction, T-AP is tumor movement in AP direction, T-CC is tumor movement in CC direction, SB is shallow breathing, DIBH is deep inspiration breath hold, DEBH is deep expiration breath hold.

B. Signal quality

A lognormal fit was used to estimate the mean value of the signal quality, which was 5.4 共standard deviation 3.0兲 for the tracking signal and 4.2 共standard deviation 3.6兲 for all signals. In all cases the AP or CC direction of the surface marker at the xiphoid process was used as tracking signal. When considering the signal quality as a function of marker position the thorax, costal arch, and xiphoid process markers tend to be superior compared to abdominal markers 共p ⬍ 0.0001兲 共see also Table II兲. No significant difference in terms of signal quality was seen between thorax and costal arch marker to xiphoid process marker. Of note is that signals in CC direction had a worse signal quality than those in AP direction 共p ⬍ 0.002兲. Medical Physics, Vol. 33, No. 8, August 2006

C. Multislice investigation: Marker-tumor correlation

The tumor characteristics were as follows for the 16 patients with target positions evenly distributed over the whole lung. The clinical target volume 共CTV兲 ranged from 1.6 to 83.9 cm3 with a mean of 18.3± 20.2 cm3. The mean forced expiratory volume 共FEV1兲 was 2.4 liter 共range 1.2 to 3.6 liter兲. The mean vital capacity 共VC兲 was 3.4 liter 共range 1.7 to 4.6 liter兲. A typical diagram of tumor and tumor marker movement is shown in Fig. 4. In seven out of 16 cases the tumor marker was also the tracking marker. In the other cases 共nine out of 16兲 the tumor marker was different from the tracking marker 共placed at xiphoid process兲. In these cases the mean Pearson

2873

Stock et al.: Real-time monitoring of external marker for gated DIBH

2873

TABLE III. Pearson correlation coefficients and slopes of linear regression of different marker and tumor direction combinations. T-CC is movement of tumor in cranial-caudal direction, T-AP is movement of tumor in anterior-posterior direction, M-CC is movement of tumor marker in cranial-caudal direction, M-AP is movement of tumor marker in anterior-posterior direction, SD is standard deviation. T-CC/ M-AP Correlation coefficient mean 0.81 SD 0.16 range 0.47–0.96 Slope mean −0.28 SD 1.13 range −2.09– 2.96

T-AP/ M-AP

T-CC/ M-CC

T-AP/ M-CC

0.89 0.11 0.64–0.99

0.77 0.20 0.12–0.94

0.84 0.20 0.24–0.99

0.81 0.77 0.13–2.97

−0.32 0.87 −2.58– 1.39

0.62 0.62 0.04–2.61

correlation between tracking marker in AP and tumor marker in AP was 0.97± 0.03. Due to this very good correlation, the two groups were evaluated as pooled data. The mean tumor movement from SB to DIBH was 13.2± 12.6 mm in CC direction. The mean in AP was 12.3± 6.9 mm. The largest amplitudes were observed in CC direction, in agreement with findings of other groups.2,14 In two patients the tumor moved in cranial direction because the lesion had contact with the pleura and performed its movement in the same direction as the rib cage. In all other patients the movement of the diaphragm was very dominant and lesions were pushed in caudal direction.

In general, good correlations of tumor marker and tumor movements were found in all directions with a mean absolute coefficient of 0.83± 0.17 共for further detail see Table III兲. For patient 1 the regression lines are shown in Fig. 5 as an example. D. Implementation into clinical routine

In a first step the feasibility of the system was tested for three lymphoma patients. If the patients were unable to hold their breath the linear accelerator was switched off manually. This was only the case in one breath hold. The second step

FIG. 5. Regression line for patient 1 for all different combinations of marker and tumor movement 共in mm兲. M-AP is marker movement in AP direction, M-CC is marker movement in CC direction, T-AP is tumor movement in AP direction, T-CC is tumor movement in CC direction.

Medical Physics, Vol. 33, No. 8, August 2006

2874

Stock et al.: Real-time monitoring of external marker for gated DIBH

will be the implementation for SBRT treatments. In order to evaluate the feasibility nine SBRT patients were asked to hold their breath as long as it was comfortable. The mean maximum deep inspiration duration was 46 s with a range of 24– 97 s. In our clinical SBRT routine the prescription dose was mostly 12.5 Gy to the 65% isodose 共encloses the target兲. In total, three fractions were given during 1 week. On average, 2157 MU per planning target volume 共PTV兲 per fraction were applied with five beams 关three beams for arc therapy and seven beams for conformal multileaf collimator 共MLC兲 technique兴. The mean duration per beam at the maximum dose rate of 400 MU/ min was 70 s 共97 s for arc therapy and 47 s for conformal MLC technique兲. Thus, on average 10–15 breath holds would be needed for one single treatment with the arc therapy technique or conformal MLC technique. The total treatment time including 1 min break between each DIBH amounts to 15– 20 min. That means only patients in good condition who are able to perform several DIBH in a row can benefit from DIBH. IV. DISCUSSION The introduced technique for gated DIBH is technically feasible, it was well tolerated by all patients, and has the potential to improve the reproducibility of the tumor position in DIBH. Therefore, the possibility to decrease margins adds to the known advantages of deep inspiration, such as increased lung volume, smaller high dose volume in the lung, decreased lung density, and tumor immobilization due to the breath hold.2,37 All these benefits enable dose escalation for targets in the lung. A. DIBH monitoring and feedback system

The hypothesis of our work was that supplying the patient with information about the actual breathing position will help to reproduce the tumor position in DIBH. One prerequisite for this approach is a good correlation between tumor movement and marker movement, which was investigated in a multislice CT study. The impact of the feedback system was shown by the increased reproducibility. For some patients we observed a good reproducibility even without feedback. Nevertheless, we think that using the system also for these patients provides a good verification tool for monitoring the DIBH. In addition, the tracking system can be used for triggering the linear accelerator. The linear accelerator can be automatically switched off when the patient’s breath hold is out of the previously defined tolerance criteria, similar to a type of technique that has already been implemented for the treatment of uveal melanoma38 in our institute. We initially chose a DIBH amplitude tolerance interval of 2 mm. In the future we aim to define this interval individually depending on patient’s ability and margin. As initial value we will set the interval to 15–20% of the DIBH amplitude, which corresponds to the measured reproducibility in this study. We addressed the usability of external markers for tracking a DIBH maneuver as a function of the position on the Medical Physics, Vol. 33, No. 8, August 2006

2874

patient surface and the direction of the movement. By analyzing the signal qualities, marker movements in AP direction produced a better signal quality than in CC direction. Averaged over all patients, the markers on the xiphoid process and thorax in AP direction showed the best quality factor and are therefore ideal for the use as tracking/gating signal. The marker positioned at the thorax showed the same performance as the xiphoid process marker. However, for some adipose patients it was not possible to use these markers because of the angular limits of the IR cameras. For markers positioned on the abdomen, the DIBH amplitude in AP direction is high, but also the SB amplitude is high and thus the signal quality is poor. In the next version of the client software an algorithm will be implemented which automatically calculates the signal quality immediately and the best marker signal for tracking can be chosen by the user. B. Marker-tumor correlation: Inter- and intrasubject variability

When looking at the standard deviation of tumor movements it is obvious that the excursion was very individual. Also, the correlation and regression slope varied with factors like tumor location 共whether the tumor has contact with the pleura兲, age, type of breathing pattern, the marker location relative to the tumor, etc.39–41 共intersubject variability兲. It is therefore advisable that slope and correlation values should be assessed individually. The age has an influence because the rib cartilage of older patients ossifies and the rib cage is not so flexible. Therefore, these patients have a stronger type of abdominal breathing. Adipose patients had strong abdominal breathing too. The two minima 共0.12 and 0.24兲 in the correlation of T-CC/ M-AP and T-CC/ M-CC were due to an adipose patient 共see Table III兲. Breathing can vary considerably between CT acquisition and treatment delivery or even during treatment 共intrasubject variability兲. As already shown by Liu et al.,40 the slope can be different from one session to another. In our opinion this influence can be minimized by using the SBF and the feedback system. Certainly, an additional margin has to be added due to the variation of the slope. To quantify this contribution more research is needed. C. Implications of treatment margins for SBRT

We used the tumor-marker correlation to assess the internal margin with/without feedback system. The internal margin accounts for interbreath hold variation and intrabreath hold variation during treatment. Additionally, we used the range of tumor movements measured from dynamic scans in SB to estimate the internal margin in a SB treatment situation for comparison. The calculations of the inter- and intrabreath hold contribution to the margin are shown in the following equations: mCC = mCC,Intra + mCC,Inter ;

冉

mCC,Intra = abs

␴Intra sT−CC,M−AP

mAP = mAP,Intra + mAP,Inter

冊

;

2875

Stock et al.: Real-time monitoring of external marker for gated DIBH

2875

TABLE IV. Internal margins in CC and AP direction during shallow breathing and DIBH with and without feedback. FB is feedback, SD is standard deviation, mCC is internal margin in CC direction, mAP is internal margin in AP direction, mCC,Inter is interbreath hold margin in CC direction, mAP,Inter is interbreath hold margin in AP direction.

Without FB mean 关mm兴 SD 关mm兴 range 关mm兴 With FB mean 关mm兴 SD 关mm兴 range 关mm兴 Shallow breathing mean 关mm兴 SD 关mm兴 range 关mm兴

冉 冉 冉

mAP,Intra = abs

mCC,Inter = abs

mAP,Inter = abs

␴Intra sT−AP,M−AP amax sT−CC,M−AP amax

冊 冊 冊

sT−AP,M−AP

mCC

mAP

mCC,Intra

mAP,Intra

mCC,Intra

mAP,Inter

7.8 6.7 1.4–21.4

5.6 4.8 1.0–20.4

0.8 0.8 0.1–2.4

0.6 0.6 0.1–2.6

7.0 6.1 1.2–19.3

5.0 4.3 0.7–17.8

4.5 4.2 0.3–12.3

2.8 2.3 0.2–8.8

0.8 0.8 0.1–2.7

0.6 0.6 0.1–2.3

3.7 3.5 0.2–9.8

2.2 1.8 0.1–6.5

6.4 6.4 0.0–20.0

3.3 2.5 0.5–10.5

;

;

,

where amax = max兵ai − aref其 " 兵ai其, m is total internal margin in CC or AP direction, ␴Intra is mean standard deviation of AP tumor marker position during breath holds, amax is maximum deviation of AP tumor marker position from reference DIBH position, sT-CC,M-AP is slope of linear fit function between tumor movement in CC and tumor marker in AP direction, and sT-AP,M-AP is slope of linear fit function between tumor movement in AP and tumor marker in AP direction. The internal margins calculated for all 16 patients are shown in Table IV. The tumor intrabreath hold stability measured using the dynamic scan data was comparable to those determined from marker tracking. The mean value in AP 0.5± 0.5 mm 共range of 0.0 to 1.4 mm兲 and in CC 0.7± 0.7 mm 共range of 0.0 to 2.7 mm兲 were about the same compared to the values presented in Table IV. There was no difference in margins for the intrabreath hold stability with and without feedback. However, a large difference can be seen in the interbreath hold variation. The mean margin for interbreath hold variability could be reduced by about 3 mm in both CC and AP. The standard deviation was rather large, and therefore the amount of margin reduction varies from patient to patient. This depends on the amplitude of tumor movement and also on the steepness of the regression line. In current clinical SBRT practice with abdominal compression, treatment margins 共internal plus setup margin兲 vary between 5 and 10 mm in CC direction and 5 to7 mm for AP and lateral direction.31–33 The main benefit of the GDIBH method as described above is a reproducible DIBH position. Medical Physics, Vol. 33, No. 8, August 2006

But, due to strong intersubject variability a patient-based margin determination and possible reduction is reasonable. Patients treated with this approach should be properly selected since not every patient benefits from GDIBH. A criterion for treatment with our technique is conceivable in a way that only tumors with movements larger than 5 mm are chosen following the recommendation of the AAPM Task Group 76 共Report of AAPM Task Group 76. The Management of Respiratory Motion in Radiation Oncology: Handout for AAPM 2005 Continuing Education Session. 2005兲. Also, adipose patients providing bad correlation data seemed to be not eligible. D. Comments on current GDIBH solution

One drawback of GDIBH based on IR tracking markers is due to the geometry of the IR cameras. In adipose patients the perspective angle of the two IR cameras is too flat and markers on the thorax cannot be detected. We recognized this problem in about 10% of our study patients. This can be overcome by placing more markers on the abdomen and the costal arch. The marker at the xiphoid process was visible for all patients. We did not use the external marker system as a positioning device, but to monitor the movement of the surface during SB and mainly from SB to DIBH. Due to the fact that the IR tracking system provides absolute positions of the marker, we had to implement an offset correction procedure in the client software. Before training, the mean marker positions during SB were calculated and applied to the marker signals. This had to be done whenever setup corrections were necessary due to the verification CT before treatment. Offset correction was carried out with caution because it has a direct impact on the amplitude between SB and DIBH. In addition, we observed a baseline drifting after performing a DIBH for most of the patients 共see Fig. 6兲. This can be explained by muscles being not fully relaxed after a DIBH. It takes several breathing cycles to relax the muscles again. Therefore, we

2876

Stock et al.: Real-time monitoring of external marker for gated DIBH

2876

FIG. 6. Baseline drifting of the feedback signal after several DIBHs. The first two DIBH’s were performed without and following three with feedback.

calculated the mean marker positions only during a fully relaxed shallow breathing. During training the feedback and especially the baseline 关see Fig. 1共b兲兴 can help the patient to get back to the previous mean SB position.

ter Winkler and Stefan Lang for fruitful discussions during the whole study. Special thanks to Mr. Peter Hwezda for constructing the arm rest of the SBF and the screen holder. K. Kontrisova was supported by the EC project HCMPT2001-0318.

E. Implementation into clinical routine a兲

In free-breathing gating the irradiation is mostly performed during the expiration phase, which is then defined as the gating window.40 This phase is in total one-quarter of the whole breathing cycle, and that means the irradiation takes four times longer than, for example, a treatment during shallow breathing with a diaphragm control. For a DIBH treatment the beams may have to be split up due to restrictions in breath hold duration. However, the treatment time will still be shorter compared to free-breathing gating. This is in line with Berson et al.,8 who found a significant decrease of treatment time by a factor of 2 with breath holding. V. CONCLUSION The results of the DIBH monitoring system based on external marker tracking, including the integrated feedback device, are very promising for the implementation of GDIBH for SBRT. The real-time system is noninvasive and simple to use. It has been already clinically applied to other tumor sites without any modifications, e.g., lymphoma treatments. After approval of the ethics committee, a clinical study will be conducted to evaluate the positive influence of GDIBH and additional margin reduction for treatment of targets in the lung. However, for GDIBH any decrease in margins needs to be performed with precaution and additional verification tools of tumor positions, e.g., with on-line or on-beam portal images, are preferred. The concept and the findings of realtime DIBH monitoring depending on thoracic movements can be transferred to a full 3D surface image guidance system.42 ACKNOWLEDGMENTS The authors would like to thank BrainLAB for implementing the DDE server in the ExacTrac® software and PeMedical Physics, Vol. 33, No. 8, August 2006

Electronic mail: [email protected] K. E. Rosenzweig et al., “The deep inspiration breath-hold technique in the treatment of inoperable non-small-cell lung cancer,” Int. J. Radiat. Oncol., Biol., Phys. 48, 81–87 共2000兲. 2 J. Hanley et al., “Deep inspiration breath-hold technique for lung tumors: the potential value of target immobilization and reduced lung density in dose escalation,” Int. J. Radiat. Oncol., Biol., Phys. 45, 603–611 共1999兲. 3 K. Nagyova et al., “Is there room for respiration controlled delivery in stereotactic body radiotherapy?” Radiother. Oncol. 76 共Supplement 2兲, S155 共2005兲. 4 J. A. Hayman et al., “Dose escalation in non-small cell lung cancer using three-dimensional conformal radiation therapy: Update of a phase I trial,” J. Clin. Oncol. 19, 127–136 共2001兲. 5 Radiation Therapy Oncology Group RTOG 0236. A. Phase I. I. Trial of Stereotactic Body Radiation Therapy 共SBRT兲 in the Treatment of Patients with Medically Inoperable Stage I/II Non-Small Cell Lung Cancer, November 2005. 6 ICRU Report 62. Prescribing, Recording, and Reporting Photon Beam Therapy 共Supplement to ICRU Report 50兲. Bethesda MD, International Commission on Radiation Units and Measurements, 1999. 7 G. S. Mageras and E. Yorke, “Deep inspiration breath hold and respiratory gating strategies for reducing organ motion in radiation treatment,” Semin. Radiat. Oncol. 14, 65–75 共2004兲. 8 A. M. Berson et al., “Clinical experience using respiratory gated radiation therapy: Comparison of free-breathing and breath-hold techniques,” Int. J. Radiat. Oncol., Biol., Phys. 60, 419–426 共2004兲. 9 H. D. Kubo et al., “Breathing-synchronized radiotherapy program at the University of California Davis Cancer Center,” Med. Phys. 27, 346–353 共2000兲. 10 G. S. Mageras et al., “Fluoroscopic evaluation of diaphragmatic motion reduction with a respiratory gated radiotherapy system,” J. Appl. Clin. Med. Phys. 2, 191–200 共2001兲. 11 S. A. Nehmeh et al., “Effect of respiratory gating on reducing lung motion artefacts in P. E. T. imaging of lung cancer,” J. Appl. Clin. Med. Phys. 2, 366–371 共2001兲. 12 H. Shirato et al., “Physical aspects of a real-time tumor-tracking system for gated radiotherapy,” Int. J. Radiat. Oncol., Biol., Phys. 48, 1187–1195 共2000兲. 13 S. Shimizu et al., “Detection of lung tumor movement in real-time tumortracking radiotherapy,” Int. J. Radiat. Oncol., Biol., Phys. 51, 304–310 共2001兲. 14 Y. Seppenwoolde et al., “Precise and real-time measurement of 3D tumor motion in lung due to breathing and heartbeat, measured during radio1

2877

Stock et al.: Real-time monitoring of external marker for gated DIBH

therapy,” Int. J. Radiat. Oncol., Biol., Phys. 53, 822–834 共2002兲. T. Zhang et al., “Application of the respiratory gated radiotherapy,” Med. Phys. 30, 3165–3171 共2003兲. 16 T. Kimura et al., “Reproducibility of organ position using voluntary breath-hold method with spirometer for extracranial stereotactic radiotherapy,” Int. J. Radiat. Oncol., Biol., Phys. 60, 1307–1313 共2004兲. 17 H. D. Kubo and B. C. Hill, “Respiration gated radiotherapy treatment: A technical study,” Phys. Med. Biol. 41, 83–91 共1996兲. 18 Y. Tsunashima et al., “Correlation between the respiratory waveform measured using a respiratory sensor and 3D tumor motion in gated radiotherapy,” Int. J. Radiat. Oncol., Biol., Phys. 60, 951–958 共2004兲. 19 J. W. Wong et al., “The use of active breathing control 共ABC兲 to reduce margin for breathing motion,” Int. J. Radiat. Oncol., Biol., Phys. 44, 911–919 共1999兲. 20 P. C. F. Cheung et al., “Reproducibility of lung tumor position and reduction of lung mass within the planning target volume using active breathing control 共ABC兲,” Int. J. Radiat. Oncol., Biol., Phys. 57, 1437–1442 共2003兲. 21 L. A. Dawson et al., “The reproducibility of organ position using active breathing control 共ABC兲 during liver radiotherapy,” Int. J. Radiat. Oncol., Biol., Phys. 51, 1410–1421 共2001兲. 22 V. M. Remouchamps et al., “Significant reductions in heart and lung doses using deep inspiration breath hold with active breathing control and intensity-modulated radiation therapy for patients treated with locoregional breast irradiation,” Int. J. Radiat. Oncol., Biol., Phys. 55, 392–406 共2003兲. 23 E. M. Wilson et al., “Validation of active breathing control in patients with non-small-cell lung cancer to be treated with chartwel,” Int. J. Radiat. Oncol., Biol., Phys. 57, 864–874 共2003兲. 24 D. Mah et al., “Technical aspects of the deep inspiration breath-hold technique in the treatment of thoracic cancer,” Int. J. Radiat. Oncol., Biol., Phys. 48, 1175–1185 共2000兲. 25 A. N. Pedersen et al., “Breathing adapted radiotherapy of breast cancer: Reduction of cardiac and pulmonary doses using voluntary inspiration breath hold,” Radiother. Oncol. 72, 53–60 共2004兲. 26 E. A. Barnes et al., “Dosimetric evaluation of lung tumor immobilization using breath hold at deep inspiration,” Int. J. Radiat. Oncol., Biol., Phys. 50, 1091–1098 共2001兲. 27 D. J. W. Kim et al., “Held-breath self-gating technique for radiotherapy of non-small-cell lung cancer: A feasibility study,” Int. J. Radiat. Oncol., Biol., Phys. 49, 43–49 共2001兲. 28 H. Onishi et al., “CT evaluation of patient deep inspiration self-breathholding: How precisely can patients reproduce the tumor position in the absence of respiratory monitoring devices?” Med. Phys. 30, 1183–1187 15

Medical Physics, Vol. 33, No. 8, August 2006

2877

共2003兲. I. Lax et al., “Stereotactic radiotherapy of malignancies in the abdomen,” Acta Oncol. 33, 677–683 共1994兲. 30 K. K. Herfarth et al., “Extracranial stereotactic radiation therapy: Set-up accuracy of patients treated for liver metastasis,” Int. J. Radiat. Oncol., Biol., Phys. 46, 329–335 共2000兲. 31 J. Wulf et al., “Stereotactic radiotherapy of extracranial targets: CTsimulation and accuracy of treatment in the stereotactic body frame,” Radiother. Oncol. 57, 225–236 共2000兲. 32 Y. Negoro et al., “The effectiveness of an immobilization device in conformal radiotherapy for lung tumor: Reduction of respiratory tumor movement and evaluation of the daily setup accuracy,” Int. J. Radiat. Oncol., Biol., Phys. 50, 889–898 共2001兲. 33 J. Wulf et al., “Stereotactic radiotherapy for primary lung cancer and pulmonary metastases: A non-invasive treatment approach in medically inoperable patients,” Int. J. Radiat. Oncol., Biol., Phys. 60, 186–196 共2004兲. 34 J. Wulf, K. Baier, G. Mueller, and M. P. Flentje, “Dose-response in stereotactic irradiation of lung tumors,” Radiother. Oncol. 77, 83–87 共2005兲. 35 R. Halperin et al., “Setup reproducibility in radiation therapy for lung cancer: A. comparison between T-Bar and expanded foam immobilization devices,” Int. J. Radiat. Oncol., Biol., Phys. 43, 211–216 共1999兲. 36 G. Soete et al., “Initial clinical experience with infrared-reflecting skin markers in the positioning of patients treated by conformal radiotherapy for prostate cancer,” Int. J. Radiat. Oncol., Biol., Phys. 52, 694–698 共2002兲. 37 E. A. Barnes et al., “Dosimetric evaluation of lung tumor immobilization using breath hold at deep inspiration,” Int. J. Radiat. Oncol., Biol., Phys. 50, 1091–1098 共2001兲. 38 B. Petersch et al., “Automatic real-time surveillance of eye position and gating for stereotactic radiotherapy of uveal melanoma,” Med. Phys. 31, 3521–3527 共2004兲. 39 N. Koch et al., “Evaluation of internal lung motion for respiratory-gated radiotherapy using MRI: Part I-Correlating internal lung motion with skin fiducial motion,” Int. J. Radiat. Oncol., Biol., Phys. 60, 1459–1472 共2004兲. 40 H. H. Liu et al., “Evaluation of internal lung motion for respiratory-gated radiotherapy using MRI: Part II-Margin reduction of internal target volume,” Int. J. Radiat. Oncol., Biol., Phys. 60, 1473–1483 共2004兲. 41 C. Plathow et al., “Influence of different breathing maneuvers on internal and external organ motion: Use of fiducial markers in dynamic MRI,” Int. J. Radiat. Oncol., Biol., Phys. 62, 238–245 共2005兲. 42 G. T. Y. Chen et al., “Clinical Implementation of IGRT,” Radiother. Oncol. 76, S10 共2005兲. 29