- Email: [email protected]
Experimental verification of a two-dimensional respiratory motion compensation system with ultrasound tracking technique in radiation therapy
Physica Medica 49 (2018) 11–18
Contents lists available at ScienceDirect
Physica Medica journal homepage: www.elsevier.com/locate/ejmp
Technical note
Experimental verification of a two-dimensional respiratory motion compensation system with ultrasound tracking technique in radiation therapy
T
⁎
Lai-Lei Tingc, Ho-Chiao Chuanga, , Ai-Ho Liaob, Chia-Chun Kuoc, Hsiao-Wei Yuc, Yi-Liang Zhoua, Der-Chi Tiena, Shiu-Chen Jengc,d, Jeng-Fong Chiouc,e,f a
Department of Mechanical Engineering National Taipei University of Technology, No. 1, Sec. 3, Chung-Hsiao E. Rd., Taipei 10608, Taiwan Graduate Institute of Biomedical Engineering, National Taiwan University of Science and Technology, Taipei 10607, Taiwan Department of Radiation Oncology, Taipei Medical University Hospital, Taipei, Taiwan No. 252, Wu Hsing Street, Taipei City 110, Taiwan d School of Dentistry, College of Oral Medicine, Taipei Medical University, Taipei, Taiwan e Department of Radiology, School of Medicine, College of Medicine, Taipei Medical University, Taipei, Taiwan f Taipei Cancer Center, Taipei Medical University, No. 252, Wu Hsing Street, Taipei City 110, Taiwan b c
A R T I C LE I N FO
A B S T R A C T
Keywords: Respiratory motion compensation Ultrasound image tracking Organ motion Planning target volume GAFchromic EBT3 film
This study proposed respiratory motion compensation system (RMCS) combined with an ultrasound image tracking algorithm (UITA) to compensate for respiration-induced tumor motion during radiotherapy, and to address the problem of inaccurate radiation dose delivery caused by respiratory movement. This study used an ultrasound imaging system to monitor respiratory movements combined with the proposed UITA and RMCS for tracking and compensation of the respiratory motion. Respiratory motion compensation was performed using prerecorded human respiratory motion signals and also sinusoidal signals. A linear accelerator was used to deliver radiation doses to GAFchromic EBT3 dosimetry film, and the conformity index (CI), rootmean-square error, compensation rate (CR), and planning target volume (PTV) were used to evaluate the tracking and compensation performance of the proposed system. Human respiratory pattern signals were captured using the UITA and compensated by the RMCS, which yielded CR values of 34–78%. In addition, the maximum coronal area of the PTV ranged from 85.53 mm2 to 351.11 mm2 (uncompensated), which reduced to from 17.72 mm2 to 66.17 mm2 after compensation, with an area reduction ratio of up to 90%. In real-time monitoring of the respiration compensation state, the CI values for 85% and 90% isodose areas increased to 0.7 and 0.68, respectively. The proposed UITA and RMCS can reduce the movement of the tracked target relative to the LINAC in radiation therapy, thereby reducing the required size of the PTV margin and increasing the effect of the radiation dose received by the treatment target.
1. Introduction Organ motion during radiotherapy as typically caused by breathing, the heartbeat, and gastrointestinal motility results in tumor motion near the chest and abdomen, (e.g., for lung, liver, and pancreatic cancers), which affects the accuracy of transmission of the radiation dose and hence the treatment effectiveness [1,2]. Furthermore, respirationinduced movement of the thoracic cage causes the tumor to move, especially in the vicinity of the diaphragm area. Studies have found that the maximum movements of the liver in the superior-inferior (SI), anterior-posterior (AP), and right-left (RL) directions were 47.4 mm,
⁎
30.2 mm, and 15.4 mm [3], respectively, with corresponding average amplitudes of lung-tumor movements of 24.6 mm, 8.2 mm, and 2.8 mm [4], respectively. The current technological approaches for managing such motion in clinical practice can be divided into three categories: motion-encompassing treatment, gating, and real-time tracking. Motion-encompassing treatment involves increasing the treatment area to ensure that the area of cancer cells is covered completely [5]; however, the associated increase in the irradiation margin is bound to cause irradiation damage to the normal tissue around the tumor. Gating involves using certain monitoring techniques to observe an organ or the movement of skin surface in real time so as to predict the location of the
Corresponding author. E-mail address: [email protected] (H.-C. Chuang).
https://doi.org/10.1016/j.ejmp.2018.04.393 Received 4 January 2018; Received in revised form 11 April 2018; Accepted 17 April 2018 1120-1797/ © 2018 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
therapeutic effects during radiotherapy were assessed by using dosimetry film (GAFchromic EBT3, ISP, Wayne, USA) to quantify the distribution of received radiation doses.
tumor. When the treatment target moves into the irradiating area, the radiation source will be activated for therapy, which should accurately deliver the radiation dose to the tumor and thereby spare the normal tissue from being irradiated [6–10]. Meyer et al. [11] proposed a method by using the TomoDirect and breath-hold 3DCRT techniques for left breast treatments, and to determine if the lack of respiratory gating is a handicap for cardiac sparing. Miura et al. [12] also reported a method of evaluating respiratory induced organ motion by vector volume histogram. They considered that it might be a useful technique in evaluating organ volumetric variation during respiration-induced organ motion. Real-time tracking technology is the best way to target tumors and improve the overall therapeutic effectiveness. This involves monitoring the target and/or tumor motions in real time using imaging systems or other monitoring devices for tracking and compensation during radiotherapy in order to accurately deliver the radiation dose [13–15]. Realtime tracking technologies can be divided into two types: direct and indirect monitoring. Direct monitoring involves observing the tumor or organ movement directly, such as by using X-ray imaging [16], magnetic resonance imaging (MRI) [17], electromagnetic field sensing (the Calypso system) [13,14], and ultrasound imaging [18–20]. Indirect monitoring involves observing the respiration-induced movement of the abdominal surface, and then determining the movement of organs or tumors according to an appropriate correlation model, such as realtime position management (RPM) [7,15] or strain-gauge sensing [21]. Many studies have focused on the development of motion management during radiotherapy, and such radiotherapy tracking systems include the Multileaf Collimator tracking system [20], a couch-based tracking system [21,22], and the CyberKnife system [23]. However, in the clinical situation it can be difficult to use only image monitoring for accurately tracking a lung tumor during radiotherapy without also implanting fiducial markers. This problem can be addressed by using the motion of other organs to estimate the target motion [24,25]. Cerviño et al. [25] used fluoroscopic images to evaluate diaphragm and tumor motions, and their results showed that the diaphragm motion was useful for predicting the motion near to a tumor in most patients, with a correlation coefficient of up to 0.94. In addition, Yang et al. [26] used MRI to monitor liver tumor and diaphragm movements, and found coefficients for the correlations between these two movements of 0.98 ± 0.02 (mean ± SD), 0.97 ± 0.02, and 0.08 ± 0.06 in the SI, AP, and medial-lateral directions, respectively, with the average errors in these three directions all being less than 3 mm. The present study proposed a method for the real-time target tracking and compensation of respiratory motion both in the SI and RL directions for application in radiotherapy. This work was also conducted in cooperation with the Department of Radiation Oncology at Taipei Medical University Hospital.
2.1. Experimental equipment This study employed the following experimental equipment: the RMCS, a respiratory motion simulation system (RMSS), motion control cards (PCI-7344, National Instruments; RT-DAC4), controlling software (VisSim), a linear accelerator (LINAC; Synergy, Elekta), and a commercial 2D ultrasound system (UF-4000, Fukuda Denshi, Tokyo, Japan). The ultrasound imaging system used in this study was coupled with a transducer probe (center frequency 3.5 MHz) and the parameters used in vivo experiments include the B-mode output pulse, the center frequency of the transducer (convex): 3.5 MHz, gain: 70 db, frame rate: 30 Hz, MI < 1.9, and maximum scanning depth: 24 cm. Moreover, the US probe was placed on the right subcostal margin between the midclavicular and anterior axillary lines, and the costal region was marked by tape to ensure that the skin landmark was in the same place. In order to have a fixed and steady angle of the probe during measurement, the probe was held by a special designed mechanical device. The RMCS is a couch tracking device and it comprised a couch, motor (BA1SGM7A-04AFA61, Yaskawa), ball screw (KK60, HIWIN), linear ball slide (EGH15CA, HIWIN), and acrylic plate. The couch is driven by the motor to offset the diaphragm motion in real-time. The software included the target-tracking UITA [27], VisSim, and LabVIEW (National Instruments, NI). The phantom was divided into two parts: diaphragm and solid water phantoms. The diaphragm phantom was constructed from a rubber belt attached to the inner wall of a basin, which simulated the real diaphragm as a tracking target during ultrasound imaging. The basin was filled with agarose and water as an ultrasound transmission medium, and also to simulate the internal tissues of the human body. An infrared ray was used to align the long axis of EBT3 radiochromic film with respect to SI direction. The EBT3 film was placed inside the solid water phantom (Solid Water®, Gammex) for analyzing the delivered radiation doses in two dimensions. The experimental setup is shown in Fig. 1a. The UITA employed in this study was based on our previously developed ultrasound image tracking method [27] (Fig. 1b and 2c). It was used to measure the grayscale pixel values in a captured ultrasound image and automatically determine the brightest pixel using a small white box in order to lock onto the target position within the image and observe its movement, which allowed us to calculate the target motion in real time. This study applied the UITA for diaphragm motion tracking in the SI and RL directions, and the captured motion signals were used as experimental breathing pattern input signals. Furthermore, during the experiments the UITA-detected dynamic image tracking signals of the diaphragm phantom (rubber belt) were also used as compensation source signals for the RMCS.
2. Materials and methods This study applied our previously developed ultrasound image tracking algorithm (UITA) for real-time diaphragm motion tracking. Any point of interest on ultrasound image could be tracked and locked by using UITA (written in Visual C). The algorithm captures the breathing signals from ultrasound images and uses the “peak and valley method” to analyze the displacement of chosen pixel. The obtained displacement signals were sent to the image motion control software (VisSim) with real-time data acquisition for driving RMCS. Detailed explanation of UITA is mentioned in our previous study [27]. In this study, diaphragm motion was used as a substitute for tumor movement, and an ultrasound imaging system was used for observing respiratory motion. In addition, a proposed two-axis respiratory motion compensation system (RMCS) was used to offset the movement of the treatment target relative to the radiation source. The effectiveness of the RMCS and UITA in respiratory motion compensation and the
2.2. Respiratory motion signals Five volunteers were recruited for capturing respiration motion signals (diaphragm movement) using the UITA. With the volunteers in a relaxed condition, an ultrasound probe was clamped to the abdomen, and the brightness and focal length of the obtained ultrasound image were adjusted to maximize the image clarity. The UITA was then used with VisSim software to capture the movement of the bottom edge of diaphragm in the SI and RL directions. To ensure consistent observations, the recorded respiration motion signals of the five volunteers and sinusoidal signals with four frequencies (0.167 Hz, 0.2 Hz, 0.25 Hz, and 0.333 Hz) and amplitudes (10 mm in the SI direction and 5 mm in the RL direction) were used as the input signals of respiratory motion.
12
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
Fig. 1. Experimental setup and respiratory motion tracking using the ultrasound image tracking algorithm (UITA). (a) Experimental setup. A, Linear accelerator; B, ultrasound system; C, respiratory motion simulation system; D, respiratory motion compensation system; E, ultrasound probe; F, solid water phantom; G, diaphragm phantom. (b) UITA tracking interface. (c) Respiratory-motion-signal capturing window.
motion more quickly and to improve the compensation effect. During the experiments, the LINAC was activated at 6 MV to apply the standard dose for detection by the EBT3 film: delivered dose = 180 cGy, dose rate = 150 MU/min, radiation field size = 3 cm × 3 cm, source-to-skin distance = 100 cm, and EBT3 insertion depth = 1.5 cm. We assessed the tracking and compensating performance of the UITA with the RMCS so as to evaluate the feasibility of using our proposed system in future clinical applications.
2.3. Experiments on respiratory motion compensation The experimental equipment (RMCS, RMSS, two phantoms, and the ultrasound imaging system) were set up on the treatment couch under the LINAC, and the radiation source was aligned with the EBT3 film inside the solid water phantom. The prerecorded respiration motion signals were input to the RMSS to drive both the solid water and diaphragm phantoms. The ultrasound imaging system was used to detect the movement of the diaphragm phantom (the rubber belt) in the SI and RL directions of the diaphragm, and send the obtained ultrasound images to the computer, while our previously developed UITA [27] was used to track the respiratory motion. The target position was acquired instantly and sent to the control program for logical operations and to output the compensation signals to the RMCS for real-time compensation. The total system delay time (350 ms) in this study prevented the RMCS from completely compensating for respiratory motion, resulting in a residual target motion defined as the difference between the RMSS and RMCS motor encoder position signals. The total delay time was measured by the recorded lag time between the two encoders of RMCS and RMSS. In this study, the RMCS-driven delay time is less than 100 ms. Thus, the proposed RMCS is able to compensate the target
2.4. Definition of tracking errors and compensation rate The root-mean-square error (RMSE) values of the uncompensated respiratory motion signals and the residual motion signals after compensation were both calculated to evaluate the system tracking error due to the delay time during the entire process of respiratory motion compensation. The compensation effect was evaluated by calculating the compensation rate (CR) defined as follows:
RMSEcom ⎞ CR (%) = ⎛1− × 100 ⎝ RMSEuncom ⎠ ⎜
⎟
(1)
where RMSEuncom is the RMSE of the uncompensated respiratory motion 13
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
residual motion signals after compensation under three different breathing modes. The figure indicates that the amplitude of the residual motion after compensation was decreased compared to that for uncompensated respiratory motion, demonstrating the good tracking performance and compensation effect of the proposed RMCS for human respiratory motion. The tracking errors (RMSE values) in the SI and RL directions of Pattern C were reduced to 2.13 mm and 0.41 mm, respectively. Table 1 lists the mean respiratory rates, RMSE, and CR for nine different breathing modes. The RMSE values were reduced due to the compensation of respiratory motion, and the tracking and compensation effects varied with the different breathing patterns (CR = 24–78%).
signals and the RMSEcom is the RMSE of the residual motion signals after compensation. 2.5. Evaluation of the planning target volume margin and reduction ratios Since the respiratory motion is the most important factor affecting the magnitude of the planning target volume (PTV), in this study the difference between the maximum and minimum peak values of the respiratory motion signal was defined as the PTV margin during a certain period of time. The area of the PTV in the coronal plane (planning target volume area [PTVA]) was the area surrounded by the PTV margin in the SI and RL directions, and the planning target volume reduction ratio (PTVRR) was calculated using PTVAuncom and PTVAcom:
3.3. Evaluation results for PTV
PTVAcom ⎞ PTVRR (%) = ⎛1− × 100 ⎝ PTVAuncom ⎠ ⎜
⎟
(2)
The observation and compensation of respiratory motion using ultrasound imaging not only reduced the relative movement between the target and the radiation source, but also corrected the baseline shift (see Fig. 2a and b), thereby reducing the PTV margin further. Table 2 lists the required PTV margin in the SI and RL directions for the nine breathing modes. The experimental results show that the PTVA (Pattern B) was reduced by a maximum of 286.69 mm2 using our proposed RMCS combined with the UITA, and that the PTVRR reached 90% (Pattern C).
2.6. Quantitation of radiation dose In order to facilitate the application of the proposed system in a clinical trial, this study used EBT3 film to verify the delivered radiation dose. In the experiments, the LINAC was activated at 6 MV to deliver the radiation dose to the EBT3 film under different conditions: static, uncompensated, and compensated. After the experiments, the EBT3 film samples receiving the radiation dose were scanned and saved as TIFF files (at a resolution of 150 dpi, with 48-bit color values and no color correction) using a verified scanner. Before quantifying the received dose on the EBT3 film and to calculate the conformity index (CI), this study first established a radiation-dose center area of 1 cm × 1 cm located at the center of field sizes of 3 cm × 3 cm and 10 cm × 10 cm, in order to obtain the average dose for both of these field sizes at the center plane (1 cm × 1 cm), which yielded a radiation dose ratio of 0.93. The TIFF images of the scanned EBT3 film samples were first analyzed using Film QA Pro software (GAFchromic), and then the optical density was converted into the radiation dose using the produced calibration curve. The data were quantified using the CI within the region of interest (ROI) [28], defined as the area on the EBT3 film where the received dose was greater than predefined values (85% and 90% isodose areas) in the static state. The CI value of the EBT3 film was calculated within these ROIs in the uncompensated and compensated states:
CI =
Areaisodose AreaROI
3.4. Analysis of EBT3 films Analyzing the optical density of the EBT3 film and converting this into the radiation dose yielded the radiation dose distribution under different testing conditions, as shown in Fig. 3a, b, and c. The analyzed 85% and 90% isodose areas in the static state were 6.24 cm2 and 5.36 cm2, respectively. The profile isodose curves at the center point of the field in the SI and RL directions after smoothing by applying the Loess method are as shown in Fig. 3d and e. Table 3 lists the quantized data (CI values) for the nine breathing modes. Compensating for respiratory movement increased the overall CI values. The increase in the CI value of Pattern D was small due to the rapid breathing pattern and small amplitude; the uncompensated CI values for 85% and 90% isodose areas were 0.59 and 0.56, respectively. Moreover, the system delay time resulted in no significant compensation effect for rapid breathing patterns; after compensation the CI values for 85% and 90% isodose areas only increased to 0.73 and 0.69, respectively. The CI values of Patterns A, B, C, and E increased more, demonstrating the excellent compensation performance of the proposed system over the human respiratory movement and increases in the 85% and 90% isodose areas within the ROI.
(3)
The general definition of CI is that a conformity index (CI) equal to 1 corresponds to ideal conformation. A CI greater than 1 indicates that the irradiated volume is greater than the target volume and includes healthy tissues. If the CI is less than 1, the target volume is only partially irradiated. In this study, A CI value of 1 indicates that the dose received within the ROI is higher than the predefined isodose area (85% or 90%), which is the best condition.
4. Discussion This study used the UITA combined with the RMCS to track and compensate for respiration-induced diaphragm motion and to increase the gradient of the profile isodose curve (Fig. 3d and 3e). The difference in the gradient of the profile isodose curve between before and after compensation was less in the RL direction (Fig. 3d), which is due to the respiration movement in that direction typically being smaller. However, compensation increased the 90% isodose area within the ROI relative to the uncompensated case. The difference in the dose gradient of the profile isodose curve was highly significant in the SI direction (Fig. 3e). Compensation resulted in obvious increases in the 90% isodose area within the ROI, with the penumbra of the dose profile (50% isodose area) almost being equal to that in the static state. In contrast, the 20% isodose area was significantly reduced after compensation compared to the uncompensated case. Therefore, the compensation of respiratory motion enhances the overall radiotherapy benefits by allowing the radiation dose to be more accurately delivered to the tumor for treatment while sparing the normal tissue from radiation damage.
3. Results 3.1. System delay time Delays were introduced during the typical respiratory motion compensation procedures, ultrasound imaging, image transmission, control program operation, and motor-driving process. The ultrasound image capture rate was 30 Hz, and the system total delay time was approximately 350 ms. 3.2. Tracking error and CR Fig. 2 shows the uncompensated respiration motion signals and 14
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
Fig. 2. Uncompensated respiration motion signals and residual motion signals after compensation. Blue solid lines are the uncompensated respiration motion signals in the right-left (RL) direction, red solid lines are the uncompensated respiration motion signals in the superior-inferior (SI) direction, green dotted lines are the residual motion signals after compensation, and Patterns B–D are the human respiratory motion signals. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) Table 1 Root-mean-square error (RMSE) and compensation rate (CR) before and after compensation. Pattern
A B C D E Sin Sin Sin Sin
#1 #2 #3 #4
Respiration frequency (Hz)
0.244 0.278 0.091 0.308 0.078 0.167 0.2 0.25 0.333
SI direction
RL direction
Uncompensated RMSE (mm)
Compensated RMSE (mm)
CR(%)
Uncompensated RMSE (mm)
Compensated RMSE (mm)
CR (%)
10.3 11.3 9.7 5.4 10.3 7.1 7.1 7.1 7.1
4.2 4.5 2.1 3.3 3.4 2.6 3.4 4.1 5.3
59 60 78 38 67 63 53 42 25
1.0 1.9 1.6 1.3 1.2 3.5 3.5 3.5 3.5
0.5 0.8 0.4 0.9 0.4 1.4 1.7 2.1 2.7
54 59 74 34 63 62 51 41 24
Abbreviations: SI, superior-inferior; RL, right-left; Patterns A–E are human respiratory motion signals; Patterns Sin #1–4 are sinusoidal signals with different frequencies. The frequency of real breathing patterns was defined as the sum of the total respiration cycles divided by total time. 15
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
Table 2 Planning target volume (PTV) margin, planning target volume area (PTVA), and planning target volume reduction ratio (PTVRR) before and after compensation. Pattern
A B C D E Sin Sin Sin Sin
#1 #2 #3 #4
SI-direction PTV margin (mm)
RL-direction PTV margin (mm)
Uncom. motion
Residual motion
Uncom. motion
Residual motion
50.5 45.5 33.1 18.9 34 20 20 20 20
24.2 20.5 9.3 12.3 23.3 8.6 11.3 13.3 17.3
4.9 7.7 5.4 4.5 3.8 10 10 10 10
2.6 3.2 1.9 3.2 2.8 4.8 5.9 6.9 8.8
Uncom. motion PTVA (mm2)
Residual motion PTVA (mm2)
PTVRR (%)
245.7 351.1 177.7 85.5 130.1 200 200 200 200
62.4 64.4 17.7 40 66.2 40.9 66 92.4 151.5
75 82 90 53 49 80 67 54 24
Abbreviation: Uncom., Uncompensated.
85 ∼ 351 mm2 to 17 ∼ 66 mm2, and the PTVRR was up to 90% after compensation (see Table 2). Although the CyberKnife [23,31] has a higher accuracy in the realtime tracking of tumor imaging, this method requires additional radiation doses. In addition, if the tumor location is difficult to observe, a marker still needs to be implanted in the patient. The present study found that the proposed ultrasound image tracking method is less accurate than the CyberKnife, but it has several advantages such as being noninvasive and not involving radiation, and allowing real-time tracking, a high image capture rate (30 Hz), and direct observations of
In this study, the tracking errors of the proposed system were all less than 5.32 mm in both the SI and RL directions. However, it is difficult to compare the measured RMSE with those found in other studies due to the use of different human respiratory pattern signals. Lee et al. [29] used Align RT to obtain three-dimensional data on the skin surface of the patient in couch-based tracking experiments. Their RMSE was 7.54 mm for a system delay time of 251 ms. Scotti et al. [30] applied the Active Breathing Coordinator device in the treatment of lung cancer, which reduced the PTV margin by 23% compared to the general treatment case. In the present study, the PTVA was reduced from
Fig. 3. Radiation dose distribution and profile under three different conditions: static, uncompensated, and compensated. Dose maps: (a) static, (b) uncompensated, and (c) compensated. Profile isodose curves: (d) RL direction and (e) SI direction. Red solid lines are the static condition, blue dashed lines are the compensated condition, and green dotted lines are the uncompensated condition. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) 16
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
observe the movements of the diaphragm and tumor movement to determine their correlation function and the curve-fitting function of the tumor motion in both the SI and RL directions. This would allow the operator to adjust the compensation gain of the RMCS, thereby increasing the accuracy of the compensation effect of the respiratory movement by using the proposed RMCS combined with UITA in clinical treatments.
Table 3 Conformity index (CI) before and after compensation. Pattern
A B C D E Sin Sin Sin Sin
#1 #2 #3 #4
85% isodose area
90% isodose area
Uncompensated
Compensated
Uncompensated
Compensated
0.11 0.08 0.21 0.59 0.28 0.31 0.34 0.36 0.34
0.72 0.70 0.81 0.73 0.77 0.86 0.80 0.70 0.55
0.00 0.00 0.1 0.56 0.04 0.23 0.27 0.27 0.28
0.69 0.68 0.75 0.69 0.69 0.85 0.79 0.68 0.49
5. Conclusions We have presented a method for the real-time tracking and compensation of respiratory motion both in the SI and RL directions based on ultrasound imaging combined with the proposed UITA and RMCS for application in radiotherapy, and have simulated clinical radiotherapy planning experiments. The experimental results have shown that the proposed method is feasible for correcting respiratory motion in radiation therapy. Ultrasound imaging is not only a noninvasive method that does not require additional radiation doses for monitoring in vivo organ movement, it also can be applied with the proposed UITA and RMCS for real-time tracking and compensation to solve the problem of respiratory movement in radiation therapy. The results also indicate that the PTV margin has been decreased after respiratory motion compensation, thereby reducing radiation damage to normal tissues. Moreover, the results further suggest that the accuracy of the dose delivery in the field size can be increased, thereby improving the efficacy of radiation therapy. However, future studies into reducing the system delay time and into the direct use of ultrasound imaging for monitoring tumor movement in real time are needed before the proposed system can be used in clinical applications.
target movements—the target motion can be monitored in real time during radiation therapy without any additional radiation burden on patients. The total system delay time (350 ms) remains the biggest limitation of our proposed system, and it is also the main reason for the occurrence of tracking errors. After compensating for respiratory motion, the input signal of Pattern Sin #4 had the largest tracking error in the SI and RL directions: RMSE was 5.32 mm and 2.7 mm, respectively, and the CR was only 24%. These results are mainly attributable to the rapid changes in the input signal and the large amplitudes for this breathing pattern, which decrease the efficacy of compensation due to the response time of UITA tracking and driving RMCS compensation not being fast enough. Our group is currently addressing this problem by investigating the use of phase-lead adaptive control. A phase-lead compensator (PLC) is currently under investigation to precompensate for the fixed system total delay time. The transfer function of the PLC was
⎛ s + 1 ⎞k ⎝s + a⎠
Acknowledgments This work was supported by the National Taipei University of Technology and Taipei Medical University Hospital under Contract USTP-NTUT-TMU-107-03. The authors would like to express their appreciation to the Taipei Medical University Hospital, Taiwan for providing the financial and facilities support for this study. The ethical approval is approved by the Taipei Medical University Hospital under the reference number: IRB 201501050.
(4)
Parameter a mainly affects the phase of the breathing signal, by changing the lead time of the output signal. However, the magnification of the output signal is also changed, resulting in different amplitudes for the input and output signals. Gain factor k is therefore used to ensure that the amplitude of the output signal matches that of the input signal. It is expected that the delay time of the system will be reduced or even eliminated in the near future, which would significantly improve the overall compensation effect and thereby allow further improvements in radiation therapy. Furthermore, this study only compensates for motion in the SI and RL without AP directions, which is an additional limitation of the study. Certain difficulties need to be overcome before ultrasound can be used to monitor tumor movements in clinical applications. Although the diaphragm movement is a good alternative to the movements of the lung and liver tumors due to their strong correlation in the SI direction, with coefficients as high as 0.98 ± 0.02 [26], the amplitude of the tumor motion is not necessarily the same as that of the diaphragm motion. Moreover, the coefficient for the correlation in the RL direction is only 0.08 ± 0.06, which might be due to the diaphragm moving less in that direction [26]. However, since a tumor may exhibit larger movements in the RL direction in some patients, compensating for respiratory motion in that direction is still needed for optimizing radiotherapy. Meanwhile, the nonlinearity and nonrepeatability of the tumor motion trajectory also need to be taken into account, and so clinical applications of the proposed system require the use of other imaging equipment for confirming organ movements. Before performing the compensation in the proposed RMCS combined with the UITA in radiation therapy in the future, it will be necessary to compare the obtained ultrasound images and correction cone-beam computed tomography images before administering treatment to a patient, in order to
References [1] Keall P, Mageras G, Balter J, et al. The management of respiratory motion in radiation oncology report of AAPM Task Group 76a). Med Phys 2006;33(10):3874–900. [2] Li Y, Ma J, Chen X, Tang F, Zhang X. 4DCT and CBCT based PTV margin in Stereotactic Body Radiotherapy(SBRT) of non-small cell lung tumor adhered to chest wall or diaphragm. Radiat Oncol 2016;11(1):152. [3] Wysocka B, Kassam Z, Lockwood G, et al. Interfraction and respiratory organ motion during conformal radiotherapy in gastric cancer. Int J Radiat Oncol Biol Phys 2010;77(1):53–9. [4] Seppenwoolde Y, Shirato H, Kitamura K, et al. 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 2002;53(4):822–34. [5] Purdy J. Current ICRU definitions of volumes: limitations and future directions. Semin Radiat Oncol 2004;14(1):27–40. [6] Shiinoki T, Kawamura S, Uehara T, et al. Quality assurance for respiratory-gated radiotherapy using the real-time tumor-tracking radiotherapy system. Int J Med Phys Clin Eng Radiat Oncol 2014;03(03):125–32. [7] Mageras Gyorke E. Deep inspiration breath hold and respiratory gating strategies for reducing organ motion in radiation treatment. Semin Radiat Oncol 2004;14(1):65–75. [8] Kurosawa Tomoyuki, Tachibana Hidenobu, Moriya Shunsuke, Miyakawa Shin, Nishio Teiji, Sato Masanori. Usefulness of a new online patient-specific quality assurance system for respiratory-gated radiotherapy. Phys Med 2017;43:63–72. [9] Meschini Giorgia, Seregni Matteo, Pella Andrea, Ciocca Mario, Fossati Piero, Valvo Francesca, et al. Evaluation of residual abdominal tumour motion in carbon ion gated treatments through respiratory motion modelling. Phys Med 2017;34:28–37. [10] Jong WL, Ung NM, Vannyat Ath, Rosenfeld AB, Wong JHD. Dosimetric evaluation near lung and soft tissue interface region during respiratory-gated and non-gated radiotherapy: a moving phantom study. Phys Med 2017;42:39–46. [11] Meyer Philippe, Niederst Claudine, Scius Maximilien, Jarnet Delphine, Dehaynin Nicolas, Gantier Matthieu, Waissi Waisse, Poulin Nicolas, et al. Is the lack of
17
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
[12]
[13]
[14] [15]
[16]
[17] [18] [19]
[20]
[21] Chuang H, Hsu H, Chiu W, Tien D, Wu R, Hsu C. Verification and compensation of respiratory motion using an ultrasound imaging system. Med Phys 2015;42(3):1193–9. [22] Ting L, Chuang H, Kuo C, et al. Tracking and compensation of respiration pattern by an automatic compensation system. Med Phys 2017;44(6):2077–95. [23] Inoue M, Shiomi H, Iwata H, et al. Development of system using beam's eye view images to measure respiratory motion tracking errors in image-guided robotic radiosurgery system. J Appl Clin Med Phys 2015;16(1):100–11. [24] Schwarz M, Teske H, Stoll M, Bendl R. Improving accuracy of markerless tracking of lung tumours in fluoroscopic video by incorporating diaphragm motion. J Phys Conf Ser 2014;489:012082. [25] Cerviño L, Chao A, Sandhu A, Jiang S. The diaphragm as an anatomic surrogate for lung tumor motion. Phys Med Biol 2009;54(11):3529–41. [26] Yang J, Cai J, Wang H, et al. Is diaphragm motion a good surrogate for liver tumor motion? Int J Radiat Oncol Biol Phys 2014;90(4):952–8. [27] Kuo C, Chuang H, Teng K, et al. An autotuning respiration compensation system based on ultrasound image tracking. J X-Ray Sci Technol 2016;24(6):875–92. [28] Fattori G, Seregni M, Pella A, et al. Real-time optical tracking for motion compensated irradiation with scanned particle beams at CNAO. Nucl Instrum Methods Phys Res Sect A-Accel Spectrom Dect Assoc Equip 2016;827:39–45. [29] Lee S, Chang K, Shim J, et al. Evaluation of mechanical accuracy for couch-based tracking system (CBTS). J Appl Clin Med Phys 2012;13(6):157–69. [30] Scotti V, Marrazzo L, Saieva C, et al. Impact of a breathing-control system on target margins and normal-tissue sparing in the treatment of lung cancer: experience at the radiotherapy unit of Florence University. Radiol Med 2013;119(1):13–9. [31] Kilby W, Dooley J, Kuduvalli G, Sayeh S, Maurer C. The CyberKnife® robotic radiosurgery system in 2010. Technol Cancer Res Treat 2010;9(5):433–52.
respiratory gating prejudicial for left breast TomoDirect treatments? Phys Med 2016;32(5):644–50. Miura Hideharu, Ozawa Shuichi, Kawabata Hideo, Doi Yoshiko, Kenjou Masahiro, Furukawa Kengo, et al. Method of evaluating respiratory induced organ motion by vector volume histogram. Phys Med 2016;32(12):1570–4. Wilbert J, Baier K, Hermann C, Flentje M, Guckenberger M. Accuracy of real-time couch tracking during 3-dimensional conformal radiation therapy, intensity modulated radiation therapy, and volumetric modulated arc therapy for prostate cancer. Int J Radiat Oncol Biol Phys 2013;85(1):237–42. Hansen R, Ravkilde T, Worm E, et al. Electromagnetic guided couch and multileaf collimator tracking on a TrueBeam accelerator. Med Phys 2016;43(5):2387–98. Lang S, Zeimetz J, Ochsner G, Schmid Daners M, Riesterer O, Klöck S. Development and evaluation of a prototype tracking system using the treatment couch. Med Phys 2014;41(2):021720. Worm E, Høyer M, Fledelius W, Nielsen J, Larsen L, Poulsen P. On-line use of threedimensional marker trajectory estimation from cone-beam computed tomography projections for precise setup in radiotherapy for targets with respiratory motion. Int J Radiat Oncol Biol Phys 2012;83(1):e145–51. Dinkel J, Hintze C, Tetzlaff R, et al. 4D-MRI analysis of lung tumor motion in patients with hemidiaphragmatic paralysis. Radiother Oncol 2009;91(3):449–54. Boussuges A, Gole Y, Blanc P. Diaphragmatic motion studied by M-mode ultrasonography. Chest 2009;135(2):391–400. Haji K, Royse A, Tharmaraj D, Haji D, Botha J, Royse C. Diaphragmatic regional displacement assessed by ultrasound and correlated to subphrenic organ movement in the critically ill patients-an observational study. J Crit Care 2015;30(2). 439.e7439.e13. Fast M, O'Shea T, Nill S, Oelfke U, Harris E. First evaluation of the feasibility of MLC tracking using ultrasound motion estimation. Med Phys 2016;43(8):4628–33.
18
Contents lists available at ScienceDirect
Physica Medica journal homepage: www.elsevier.com/locate/ejmp
Technical note
Experimental verification of a two-dimensional respiratory motion compensation system with ultrasound tracking technique in radiation therapy
T
⁎
Lai-Lei Tingc, Ho-Chiao Chuanga, , Ai-Ho Liaob, Chia-Chun Kuoc, Hsiao-Wei Yuc, Yi-Liang Zhoua, Der-Chi Tiena, Shiu-Chen Jengc,d, Jeng-Fong Chiouc,e,f a
Department of Mechanical Engineering National Taipei University of Technology, No. 1, Sec. 3, Chung-Hsiao E. Rd., Taipei 10608, Taiwan Graduate Institute of Biomedical Engineering, National Taiwan University of Science and Technology, Taipei 10607, Taiwan Department of Radiation Oncology, Taipei Medical University Hospital, Taipei, Taiwan No. 252, Wu Hsing Street, Taipei City 110, Taiwan d School of Dentistry, College of Oral Medicine, Taipei Medical University, Taipei, Taiwan e Department of Radiology, School of Medicine, College of Medicine, Taipei Medical University, Taipei, Taiwan f Taipei Cancer Center, Taipei Medical University, No. 252, Wu Hsing Street, Taipei City 110, Taiwan b c
A R T I C LE I N FO
A B S T R A C T
Keywords: Respiratory motion compensation Ultrasound image tracking Organ motion Planning target volume GAFchromic EBT3 film
This study proposed respiratory motion compensation system (RMCS) combined with an ultrasound image tracking algorithm (UITA) to compensate for respiration-induced tumor motion during radiotherapy, and to address the problem of inaccurate radiation dose delivery caused by respiratory movement. This study used an ultrasound imaging system to monitor respiratory movements combined with the proposed UITA and RMCS for tracking and compensation of the respiratory motion. Respiratory motion compensation was performed using prerecorded human respiratory motion signals and also sinusoidal signals. A linear accelerator was used to deliver radiation doses to GAFchromic EBT3 dosimetry film, and the conformity index (CI), rootmean-square error, compensation rate (CR), and planning target volume (PTV) were used to evaluate the tracking and compensation performance of the proposed system. Human respiratory pattern signals were captured using the UITA and compensated by the RMCS, which yielded CR values of 34–78%. In addition, the maximum coronal area of the PTV ranged from 85.53 mm2 to 351.11 mm2 (uncompensated), which reduced to from 17.72 mm2 to 66.17 mm2 after compensation, with an area reduction ratio of up to 90%. In real-time monitoring of the respiration compensation state, the CI values for 85% and 90% isodose areas increased to 0.7 and 0.68, respectively. The proposed UITA and RMCS can reduce the movement of the tracked target relative to the LINAC in radiation therapy, thereby reducing the required size of the PTV margin and increasing the effect of the radiation dose received by the treatment target.
1. Introduction Organ motion during radiotherapy as typically caused by breathing, the heartbeat, and gastrointestinal motility results in tumor motion near the chest and abdomen, (e.g., for lung, liver, and pancreatic cancers), which affects the accuracy of transmission of the radiation dose and hence the treatment effectiveness [1,2]. Furthermore, respirationinduced movement of the thoracic cage causes the tumor to move, especially in the vicinity of the diaphragm area. Studies have found that the maximum movements of the liver in the superior-inferior (SI), anterior-posterior (AP), and right-left (RL) directions were 47.4 mm,
⁎
30.2 mm, and 15.4 mm [3], respectively, with corresponding average amplitudes of lung-tumor movements of 24.6 mm, 8.2 mm, and 2.8 mm [4], respectively. The current technological approaches for managing such motion in clinical practice can be divided into three categories: motion-encompassing treatment, gating, and real-time tracking. Motion-encompassing treatment involves increasing the treatment area to ensure that the area of cancer cells is covered completely [5]; however, the associated increase in the irradiation margin is bound to cause irradiation damage to the normal tissue around the tumor. Gating involves using certain monitoring techniques to observe an organ or the movement of skin surface in real time so as to predict the location of the
Corresponding author. E-mail address: [email protected] (H.-C. Chuang).
https://doi.org/10.1016/j.ejmp.2018.04.393 Received 4 January 2018; Received in revised form 11 April 2018; Accepted 17 April 2018 1120-1797/ © 2018 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
therapeutic effects during radiotherapy were assessed by using dosimetry film (GAFchromic EBT3, ISP, Wayne, USA) to quantify the distribution of received radiation doses.
tumor. When the treatment target moves into the irradiating area, the radiation source will be activated for therapy, which should accurately deliver the radiation dose to the tumor and thereby spare the normal tissue from being irradiated [6–10]. Meyer et al. [11] proposed a method by using the TomoDirect and breath-hold 3DCRT techniques for left breast treatments, and to determine if the lack of respiratory gating is a handicap for cardiac sparing. Miura et al. [12] also reported a method of evaluating respiratory induced organ motion by vector volume histogram. They considered that it might be a useful technique in evaluating organ volumetric variation during respiration-induced organ motion. Real-time tracking technology is the best way to target tumors and improve the overall therapeutic effectiveness. This involves monitoring the target and/or tumor motions in real time using imaging systems or other monitoring devices for tracking and compensation during radiotherapy in order to accurately deliver the radiation dose [13–15]. Realtime tracking technologies can be divided into two types: direct and indirect monitoring. Direct monitoring involves observing the tumor or organ movement directly, such as by using X-ray imaging [16], magnetic resonance imaging (MRI) [17], electromagnetic field sensing (the Calypso system) [13,14], and ultrasound imaging [18–20]. Indirect monitoring involves observing the respiration-induced movement of the abdominal surface, and then determining the movement of organs or tumors according to an appropriate correlation model, such as realtime position management (RPM) [7,15] or strain-gauge sensing [21]. Many studies have focused on the development of motion management during radiotherapy, and such radiotherapy tracking systems include the Multileaf Collimator tracking system [20], a couch-based tracking system [21,22], and the CyberKnife system [23]. However, in the clinical situation it can be difficult to use only image monitoring for accurately tracking a lung tumor during radiotherapy without also implanting fiducial markers. This problem can be addressed by using the motion of other organs to estimate the target motion [24,25]. Cerviño et al. [25] used fluoroscopic images to evaluate diaphragm and tumor motions, and their results showed that the diaphragm motion was useful for predicting the motion near to a tumor in most patients, with a correlation coefficient of up to 0.94. In addition, Yang et al. [26] used MRI to monitor liver tumor and diaphragm movements, and found coefficients for the correlations between these two movements of 0.98 ± 0.02 (mean ± SD), 0.97 ± 0.02, and 0.08 ± 0.06 in the SI, AP, and medial-lateral directions, respectively, with the average errors in these three directions all being less than 3 mm. The present study proposed a method for the real-time target tracking and compensation of respiratory motion both in the SI and RL directions for application in radiotherapy. This work was also conducted in cooperation with the Department of Radiation Oncology at Taipei Medical University Hospital.
2.1. Experimental equipment This study employed the following experimental equipment: the RMCS, a respiratory motion simulation system (RMSS), motion control cards (PCI-7344, National Instruments; RT-DAC4), controlling software (VisSim), a linear accelerator (LINAC; Synergy, Elekta), and a commercial 2D ultrasound system (UF-4000, Fukuda Denshi, Tokyo, Japan). The ultrasound imaging system used in this study was coupled with a transducer probe (center frequency 3.5 MHz) and the parameters used in vivo experiments include the B-mode output pulse, the center frequency of the transducer (convex): 3.5 MHz, gain: 70 db, frame rate: 30 Hz, MI < 1.9, and maximum scanning depth: 24 cm. Moreover, the US probe was placed on the right subcostal margin between the midclavicular and anterior axillary lines, and the costal region was marked by tape to ensure that the skin landmark was in the same place. In order to have a fixed and steady angle of the probe during measurement, the probe was held by a special designed mechanical device. The RMCS is a couch tracking device and it comprised a couch, motor (BA1SGM7A-04AFA61, Yaskawa), ball screw (KK60, HIWIN), linear ball slide (EGH15CA, HIWIN), and acrylic plate. The couch is driven by the motor to offset the diaphragm motion in real-time. The software included the target-tracking UITA [27], VisSim, and LabVIEW (National Instruments, NI). The phantom was divided into two parts: diaphragm and solid water phantoms. The diaphragm phantom was constructed from a rubber belt attached to the inner wall of a basin, which simulated the real diaphragm as a tracking target during ultrasound imaging. The basin was filled with agarose and water as an ultrasound transmission medium, and also to simulate the internal tissues of the human body. An infrared ray was used to align the long axis of EBT3 radiochromic film with respect to SI direction. The EBT3 film was placed inside the solid water phantom (Solid Water®, Gammex) for analyzing the delivered radiation doses in two dimensions. The experimental setup is shown in Fig. 1a. The UITA employed in this study was based on our previously developed ultrasound image tracking method [27] (Fig. 1b and 2c). It was used to measure the grayscale pixel values in a captured ultrasound image and automatically determine the brightest pixel using a small white box in order to lock onto the target position within the image and observe its movement, which allowed us to calculate the target motion in real time. This study applied the UITA for diaphragm motion tracking in the SI and RL directions, and the captured motion signals were used as experimental breathing pattern input signals. Furthermore, during the experiments the UITA-detected dynamic image tracking signals of the diaphragm phantom (rubber belt) were also used as compensation source signals for the RMCS.
2. Materials and methods This study applied our previously developed ultrasound image tracking algorithm (UITA) for real-time diaphragm motion tracking. Any point of interest on ultrasound image could be tracked and locked by using UITA (written in Visual C). The algorithm captures the breathing signals from ultrasound images and uses the “peak and valley method” to analyze the displacement of chosen pixel. The obtained displacement signals were sent to the image motion control software (VisSim) with real-time data acquisition for driving RMCS. Detailed explanation of UITA is mentioned in our previous study [27]. In this study, diaphragm motion was used as a substitute for tumor movement, and an ultrasound imaging system was used for observing respiratory motion. In addition, a proposed two-axis respiratory motion compensation system (RMCS) was used to offset the movement of the treatment target relative to the radiation source. The effectiveness of the RMCS and UITA in respiratory motion compensation and the
2.2. Respiratory motion signals Five volunteers were recruited for capturing respiration motion signals (diaphragm movement) using the UITA. With the volunteers in a relaxed condition, an ultrasound probe was clamped to the abdomen, and the brightness and focal length of the obtained ultrasound image were adjusted to maximize the image clarity. The UITA was then used with VisSim software to capture the movement of the bottom edge of diaphragm in the SI and RL directions. To ensure consistent observations, the recorded respiration motion signals of the five volunteers and sinusoidal signals with four frequencies (0.167 Hz, 0.2 Hz, 0.25 Hz, and 0.333 Hz) and amplitudes (10 mm in the SI direction and 5 mm in the RL direction) were used as the input signals of respiratory motion.
12
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
Fig. 1. Experimental setup and respiratory motion tracking using the ultrasound image tracking algorithm (UITA). (a) Experimental setup. A, Linear accelerator; B, ultrasound system; C, respiratory motion simulation system; D, respiratory motion compensation system; E, ultrasound probe; F, solid water phantom; G, diaphragm phantom. (b) UITA tracking interface. (c) Respiratory-motion-signal capturing window.
motion more quickly and to improve the compensation effect. During the experiments, the LINAC was activated at 6 MV to apply the standard dose for detection by the EBT3 film: delivered dose = 180 cGy, dose rate = 150 MU/min, radiation field size = 3 cm × 3 cm, source-to-skin distance = 100 cm, and EBT3 insertion depth = 1.5 cm. We assessed the tracking and compensating performance of the UITA with the RMCS so as to evaluate the feasibility of using our proposed system in future clinical applications.
2.3. Experiments on respiratory motion compensation The experimental equipment (RMCS, RMSS, two phantoms, and the ultrasound imaging system) were set up on the treatment couch under the LINAC, and the radiation source was aligned with the EBT3 film inside the solid water phantom. The prerecorded respiration motion signals were input to the RMSS to drive both the solid water and diaphragm phantoms. The ultrasound imaging system was used to detect the movement of the diaphragm phantom (the rubber belt) in the SI and RL directions of the diaphragm, and send the obtained ultrasound images to the computer, while our previously developed UITA [27] was used to track the respiratory motion. The target position was acquired instantly and sent to the control program for logical operations and to output the compensation signals to the RMCS for real-time compensation. The total system delay time (350 ms) in this study prevented the RMCS from completely compensating for respiratory motion, resulting in a residual target motion defined as the difference between the RMSS and RMCS motor encoder position signals. The total delay time was measured by the recorded lag time between the two encoders of RMCS and RMSS. In this study, the RMCS-driven delay time is less than 100 ms. Thus, the proposed RMCS is able to compensate the target
2.4. Definition of tracking errors and compensation rate The root-mean-square error (RMSE) values of the uncompensated respiratory motion signals and the residual motion signals after compensation were both calculated to evaluate the system tracking error due to the delay time during the entire process of respiratory motion compensation. The compensation effect was evaluated by calculating the compensation rate (CR) defined as follows:
RMSEcom ⎞ CR (%) = ⎛1− × 100 ⎝ RMSEuncom ⎠ ⎜
⎟
(1)
where RMSEuncom is the RMSE of the uncompensated respiratory motion 13
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
residual motion signals after compensation under three different breathing modes. The figure indicates that the amplitude of the residual motion after compensation was decreased compared to that for uncompensated respiratory motion, demonstrating the good tracking performance and compensation effect of the proposed RMCS for human respiratory motion. The tracking errors (RMSE values) in the SI and RL directions of Pattern C were reduced to 2.13 mm and 0.41 mm, respectively. Table 1 lists the mean respiratory rates, RMSE, and CR for nine different breathing modes. The RMSE values were reduced due to the compensation of respiratory motion, and the tracking and compensation effects varied with the different breathing patterns (CR = 24–78%).
signals and the RMSEcom is the RMSE of the residual motion signals after compensation. 2.5. Evaluation of the planning target volume margin and reduction ratios Since the respiratory motion is the most important factor affecting the magnitude of the planning target volume (PTV), in this study the difference between the maximum and minimum peak values of the respiratory motion signal was defined as the PTV margin during a certain period of time. The area of the PTV in the coronal plane (planning target volume area [PTVA]) was the area surrounded by the PTV margin in the SI and RL directions, and the planning target volume reduction ratio (PTVRR) was calculated using PTVAuncom and PTVAcom:
3.3. Evaluation results for PTV
PTVAcom ⎞ PTVRR (%) = ⎛1− × 100 ⎝ PTVAuncom ⎠ ⎜
⎟
(2)
The observation and compensation of respiratory motion using ultrasound imaging not only reduced the relative movement between the target and the radiation source, but also corrected the baseline shift (see Fig. 2a and b), thereby reducing the PTV margin further. Table 2 lists the required PTV margin in the SI and RL directions for the nine breathing modes. The experimental results show that the PTVA (Pattern B) was reduced by a maximum of 286.69 mm2 using our proposed RMCS combined with the UITA, and that the PTVRR reached 90% (Pattern C).
2.6. Quantitation of radiation dose In order to facilitate the application of the proposed system in a clinical trial, this study used EBT3 film to verify the delivered radiation dose. In the experiments, the LINAC was activated at 6 MV to deliver the radiation dose to the EBT3 film under different conditions: static, uncompensated, and compensated. After the experiments, the EBT3 film samples receiving the radiation dose were scanned and saved as TIFF files (at a resolution of 150 dpi, with 48-bit color values and no color correction) using a verified scanner. Before quantifying the received dose on the EBT3 film and to calculate the conformity index (CI), this study first established a radiation-dose center area of 1 cm × 1 cm located at the center of field sizes of 3 cm × 3 cm and 10 cm × 10 cm, in order to obtain the average dose for both of these field sizes at the center plane (1 cm × 1 cm), which yielded a radiation dose ratio of 0.93. The TIFF images of the scanned EBT3 film samples were first analyzed using Film QA Pro software (GAFchromic), and then the optical density was converted into the radiation dose using the produced calibration curve. The data were quantified using the CI within the region of interest (ROI) [28], defined as the area on the EBT3 film where the received dose was greater than predefined values (85% and 90% isodose areas) in the static state. The CI value of the EBT3 film was calculated within these ROIs in the uncompensated and compensated states:
CI =
Areaisodose AreaROI
3.4. Analysis of EBT3 films Analyzing the optical density of the EBT3 film and converting this into the radiation dose yielded the radiation dose distribution under different testing conditions, as shown in Fig. 3a, b, and c. The analyzed 85% and 90% isodose areas in the static state were 6.24 cm2 and 5.36 cm2, respectively. The profile isodose curves at the center point of the field in the SI and RL directions after smoothing by applying the Loess method are as shown in Fig. 3d and e. Table 3 lists the quantized data (CI values) for the nine breathing modes. Compensating for respiratory movement increased the overall CI values. The increase in the CI value of Pattern D was small due to the rapid breathing pattern and small amplitude; the uncompensated CI values for 85% and 90% isodose areas were 0.59 and 0.56, respectively. Moreover, the system delay time resulted in no significant compensation effect for rapid breathing patterns; after compensation the CI values for 85% and 90% isodose areas only increased to 0.73 and 0.69, respectively. The CI values of Patterns A, B, C, and E increased more, demonstrating the excellent compensation performance of the proposed system over the human respiratory movement and increases in the 85% and 90% isodose areas within the ROI.
(3)
The general definition of CI is that a conformity index (CI) equal to 1 corresponds to ideal conformation. A CI greater than 1 indicates that the irradiated volume is greater than the target volume and includes healthy tissues. If the CI is less than 1, the target volume is only partially irradiated. In this study, A CI value of 1 indicates that the dose received within the ROI is higher than the predefined isodose area (85% or 90%), which is the best condition.
4. Discussion This study used the UITA combined with the RMCS to track and compensate for respiration-induced diaphragm motion and to increase the gradient of the profile isodose curve (Fig. 3d and 3e). The difference in the gradient of the profile isodose curve between before and after compensation was less in the RL direction (Fig. 3d), which is due to the respiration movement in that direction typically being smaller. However, compensation increased the 90% isodose area within the ROI relative to the uncompensated case. The difference in the dose gradient of the profile isodose curve was highly significant in the SI direction (Fig. 3e). Compensation resulted in obvious increases in the 90% isodose area within the ROI, with the penumbra of the dose profile (50% isodose area) almost being equal to that in the static state. In contrast, the 20% isodose area was significantly reduced after compensation compared to the uncompensated case. Therefore, the compensation of respiratory motion enhances the overall radiotherapy benefits by allowing the radiation dose to be more accurately delivered to the tumor for treatment while sparing the normal tissue from radiation damage.
3. Results 3.1. System delay time Delays were introduced during the typical respiratory motion compensation procedures, ultrasound imaging, image transmission, control program operation, and motor-driving process. The ultrasound image capture rate was 30 Hz, and the system total delay time was approximately 350 ms. 3.2. Tracking error and CR Fig. 2 shows the uncompensated respiration motion signals and 14
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
Fig. 2. Uncompensated respiration motion signals and residual motion signals after compensation. Blue solid lines are the uncompensated respiration motion signals in the right-left (RL) direction, red solid lines are the uncompensated respiration motion signals in the superior-inferior (SI) direction, green dotted lines are the residual motion signals after compensation, and Patterns B–D are the human respiratory motion signals. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) Table 1 Root-mean-square error (RMSE) and compensation rate (CR) before and after compensation. Pattern
A B C D E Sin Sin Sin Sin
#1 #2 #3 #4
Respiration frequency (Hz)
0.244 0.278 0.091 0.308 0.078 0.167 0.2 0.25 0.333
SI direction
RL direction
Uncompensated RMSE (mm)
Compensated RMSE (mm)
CR(%)
Uncompensated RMSE (mm)
Compensated RMSE (mm)
CR (%)
10.3 11.3 9.7 5.4 10.3 7.1 7.1 7.1 7.1
4.2 4.5 2.1 3.3 3.4 2.6 3.4 4.1 5.3
59 60 78 38 67 63 53 42 25
1.0 1.9 1.6 1.3 1.2 3.5 3.5 3.5 3.5
0.5 0.8 0.4 0.9 0.4 1.4 1.7 2.1 2.7
54 59 74 34 63 62 51 41 24
Abbreviations: SI, superior-inferior; RL, right-left; Patterns A–E are human respiratory motion signals; Patterns Sin #1–4 are sinusoidal signals with different frequencies. The frequency of real breathing patterns was defined as the sum of the total respiration cycles divided by total time. 15
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
Table 2 Planning target volume (PTV) margin, planning target volume area (PTVA), and planning target volume reduction ratio (PTVRR) before and after compensation. Pattern
A B C D E Sin Sin Sin Sin
#1 #2 #3 #4
SI-direction PTV margin (mm)
RL-direction PTV margin (mm)
Uncom. motion
Residual motion
Uncom. motion
Residual motion
50.5 45.5 33.1 18.9 34 20 20 20 20
24.2 20.5 9.3 12.3 23.3 8.6 11.3 13.3 17.3
4.9 7.7 5.4 4.5 3.8 10 10 10 10
2.6 3.2 1.9 3.2 2.8 4.8 5.9 6.9 8.8
Uncom. motion PTVA (mm2)
Residual motion PTVA (mm2)
PTVRR (%)
245.7 351.1 177.7 85.5 130.1 200 200 200 200
62.4 64.4 17.7 40 66.2 40.9 66 92.4 151.5
75 82 90 53 49 80 67 54 24
Abbreviation: Uncom., Uncompensated.
85 ∼ 351 mm2 to 17 ∼ 66 mm2, and the PTVRR was up to 90% after compensation (see Table 2). Although the CyberKnife [23,31] has a higher accuracy in the realtime tracking of tumor imaging, this method requires additional radiation doses. In addition, if the tumor location is difficult to observe, a marker still needs to be implanted in the patient. The present study found that the proposed ultrasound image tracking method is less accurate than the CyberKnife, but it has several advantages such as being noninvasive and not involving radiation, and allowing real-time tracking, a high image capture rate (30 Hz), and direct observations of
In this study, the tracking errors of the proposed system were all less than 5.32 mm in both the SI and RL directions. However, it is difficult to compare the measured RMSE with those found in other studies due to the use of different human respiratory pattern signals. Lee et al. [29] used Align RT to obtain three-dimensional data on the skin surface of the patient in couch-based tracking experiments. Their RMSE was 7.54 mm for a system delay time of 251 ms. Scotti et al. [30] applied the Active Breathing Coordinator device in the treatment of lung cancer, which reduced the PTV margin by 23% compared to the general treatment case. In the present study, the PTVA was reduced from
Fig. 3. Radiation dose distribution and profile under three different conditions: static, uncompensated, and compensated. Dose maps: (a) static, (b) uncompensated, and (c) compensated. Profile isodose curves: (d) RL direction and (e) SI direction. Red solid lines are the static condition, blue dashed lines are the compensated condition, and green dotted lines are the uncompensated condition. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) 16
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
observe the movements of the diaphragm and tumor movement to determine their correlation function and the curve-fitting function of the tumor motion in both the SI and RL directions. This would allow the operator to adjust the compensation gain of the RMCS, thereby increasing the accuracy of the compensation effect of the respiratory movement by using the proposed RMCS combined with UITA in clinical treatments.
Table 3 Conformity index (CI) before and after compensation. Pattern
A B C D E Sin Sin Sin Sin
#1 #2 #3 #4
85% isodose area
90% isodose area
Uncompensated
Compensated
Uncompensated
Compensated
0.11 0.08 0.21 0.59 0.28 0.31 0.34 0.36 0.34
0.72 0.70 0.81 0.73 0.77 0.86 0.80 0.70 0.55
0.00 0.00 0.1 0.56 0.04 0.23 0.27 0.27 0.28
0.69 0.68 0.75 0.69 0.69 0.85 0.79 0.68 0.49
5. Conclusions We have presented a method for the real-time tracking and compensation of respiratory motion both in the SI and RL directions based on ultrasound imaging combined with the proposed UITA and RMCS for application in radiotherapy, and have simulated clinical radiotherapy planning experiments. The experimental results have shown that the proposed method is feasible for correcting respiratory motion in radiation therapy. Ultrasound imaging is not only a noninvasive method that does not require additional radiation doses for monitoring in vivo organ movement, it also can be applied with the proposed UITA and RMCS for real-time tracking and compensation to solve the problem of respiratory movement in radiation therapy. The results also indicate that the PTV margin has been decreased after respiratory motion compensation, thereby reducing radiation damage to normal tissues. Moreover, the results further suggest that the accuracy of the dose delivery in the field size can be increased, thereby improving the efficacy of radiation therapy. However, future studies into reducing the system delay time and into the direct use of ultrasound imaging for monitoring tumor movement in real time are needed before the proposed system can be used in clinical applications.
target movements—the target motion can be monitored in real time during radiation therapy without any additional radiation burden on patients. The total system delay time (350 ms) remains the biggest limitation of our proposed system, and it is also the main reason for the occurrence of tracking errors. After compensating for respiratory motion, the input signal of Pattern Sin #4 had the largest tracking error in the SI and RL directions: RMSE was 5.32 mm and 2.7 mm, respectively, and the CR was only 24%. These results are mainly attributable to the rapid changes in the input signal and the large amplitudes for this breathing pattern, which decrease the efficacy of compensation due to the response time of UITA tracking and driving RMCS compensation not being fast enough. Our group is currently addressing this problem by investigating the use of phase-lead adaptive control. A phase-lead compensator (PLC) is currently under investigation to precompensate for the fixed system total delay time. The transfer function of the PLC was
⎛ s + 1 ⎞k ⎝s + a⎠
Acknowledgments This work was supported by the National Taipei University of Technology and Taipei Medical University Hospital under Contract USTP-NTUT-TMU-107-03. The authors would like to express their appreciation to the Taipei Medical University Hospital, Taiwan for providing the financial and facilities support for this study. The ethical approval is approved by the Taipei Medical University Hospital under the reference number: IRB 201501050.
(4)
Parameter a mainly affects the phase of the breathing signal, by changing the lead time of the output signal. However, the magnification of the output signal is also changed, resulting in different amplitudes for the input and output signals. Gain factor k is therefore used to ensure that the amplitude of the output signal matches that of the input signal. It is expected that the delay time of the system will be reduced or even eliminated in the near future, which would significantly improve the overall compensation effect and thereby allow further improvements in radiation therapy. Furthermore, this study only compensates for motion in the SI and RL without AP directions, which is an additional limitation of the study. Certain difficulties need to be overcome before ultrasound can be used to monitor tumor movements in clinical applications. Although the diaphragm movement is a good alternative to the movements of the lung and liver tumors due to their strong correlation in the SI direction, with coefficients as high as 0.98 ± 0.02 [26], the amplitude of the tumor motion is not necessarily the same as that of the diaphragm motion. Moreover, the coefficient for the correlation in the RL direction is only 0.08 ± 0.06, which might be due to the diaphragm moving less in that direction [26]. However, since a tumor may exhibit larger movements in the RL direction in some patients, compensating for respiratory motion in that direction is still needed for optimizing radiotherapy. Meanwhile, the nonlinearity and nonrepeatability of the tumor motion trajectory also need to be taken into account, and so clinical applications of the proposed system require the use of other imaging equipment for confirming organ movements. Before performing the compensation in the proposed RMCS combined with the UITA in radiation therapy in the future, it will be necessary to compare the obtained ultrasound images and correction cone-beam computed tomography images before administering treatment to a patient, in order to
References [1] Keall P, Mageras G, Balter J, et al. The management of respiratory motion in radiation oncology report of AAPM Task Group 76a). Med Phys 2006;33(10):3874–900. [2] Li Y, Ma J, Chen X, Tang F, Zhang X. 4DCT and CBCT based PTV margin in Stereotactic Body Radiotherapy(SBRT) of non-small cell lung tumor adhered to chest wall or diaphragm. Radiat Oncol 2016;11(1):152. [3] Wysocka B, Kassam Z, Lockwood G, et al. Interfraction and respiratory organ motion during conformal radiotherapy in gastric cancer. Int J Radiat Oncol Biol Phys 2010;77(1):53–9. [4] Seppenwoolde Y, Shirato H, Kitamura K, et al. 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 2002;53(4):822–34. [5] Purdy J. Current ICRU definitions of volumes: limitations and future directions. Semin Radiat Oncol 2004;14(1):27–40. [6] Shiinoki T, Kawamura S, Uehara T, et al. Quality assurance for respiratory-gated radiotherapy using the real-time tumor-tracking radiotherapy system. Int J Med Phys Clin Eng Radiat Oncol 2014;03(03):125–32. [7] Mageras Gyorke E. Deep inspiration breath hold and respiratory gating strategies for reducing organ motion in radiation treatment. Semin Radiat Oncol 2004;14(1):65–75. [8] Kurosawa Tomoyuki, Tachibana Hidenobu, Moriya Shunsuke, Miyakawa Shin, Nishio Teiji, Sato Masanori. Usefulness of a new online patient-specific quality assurance system for respiratory-gated radiotherapy. Phys Med 2017;43:63–72. [9] Meschini Giorgia, Seregni Matteo, Pella Andrea, Ciocca Mario, Fossati Piero, Valvo Francesca, et al. Evaluation of residual abdominal tumour motion in carbon ion gated treatments through respiratory motion modelling. Phys Med 2017;34:28–37. [10] Jong WL, Ung NM, Vannyat Ath, Rosenfeld AB, Wong JHD. Dosimetric evaluation near lung and soft tissue interface region during respiratory-gated and non-gated radiotherapy: a moving phantom study. Phys Med 2017;42:39–46. [11] Meyer Philippe, Niederst Claudine, Scius Maximilien, Jarnet Delphine, Dehaynin Nicolas, Gantier Matthieu, Waissi Waisse, Poulin Nicolas, et al. Is the lack of
17
Physica Medica 49 (2018) 11–18
L.-L. Ting et al.
[12]
[13]
[14] [15]
[16]
[17] [18] [19]
[20]
[21] Chuang H, Hsu H, Chiu W, Tien D, Wu R, Hsu C. Verification and compensation of respiratory motion using an ultrasound imaging system. Med Phys 2015;42(3):1193–9. [22] Ting L, Chuang H, Kuo C, et al. Tracking and compensation of respiration pattern by an automatic compensation system. Med Phys 2017;44(6):2077–95. [23] Inoue M, Shiomi H, Iwata H, et al. Development of system using beam's eye view images to measure respiratory motion tracking errors in image-guided robotic radiosurgery system. J Appl Clin Med Phys 2015;16(1):100–11. [24] Schwarz M, Teske H, Stoll M, Bendl R. Improving accuracy of markerless tracking of lung tumours in fluoroscopic video by incorporating diaphragm motion. J Phys Conf Ser 2014;489:012082. [25] Cerviño L, Chao A, Sandhu A, Jiang S. The diaphragm as an anatomic surrogate for lung tumor motion. Phys Med Biol 2009;54(11):3529–41. [26] Yang J, Cai J, Wang H, et al. Is diaphragm motion a good surrogate for liver tumor motion? Int J Radiat Oncol Biol Phys 2014;90(4):952–8. [27] Kuo C, Chuang H, Teng K, et al. An autotuning respiration compensation system based on ultrasound image tracking. J X-Ray Sci Technol 2016;24(6):875–92. [28] Fattori G, Seregni M, Pella A, et al. Real-time optical tracking for motion compensated irradiation with scanned particle beams at CNAO. Nucl Instrum Methods Phys Res Sect A-Accel Spectrom Dect Assoc Equip 2016;827:39–45. [29] Lee S, Chang K, Shim J, et al. Evaluation of mechanical accuracy for couch-based tracking system (CBTS). J Appl Clin Med Phys 2012;13(6):157–69. [30] Scotti V, Marrazzo L, Saieva C, et al. Impact of a breathing-control system on target margins and normal-tissue sparing in the treatment of lung cancer: experience at the radiotherapy unit of Florence University. Radiol Med 2013;119(1):13–9. [31] Kilby W, Dooley J, Kuduvalli G, Sayeh S, Maurer C. The CyberKnife® robotic radiosurgery system in 2010. Technol Cancer Res Treat 2010;9(5):433–52.
respiratory gating prejudicial for left breast TomoDirect treatments? Phys Med 2016;32(5):644–50. Miura Hideharu, Ozawa Shuichi, Kawabata Hideo, Doi Yoshiko, Kenjou Masahiro, Furukawa Kengo, et al. Method of evaluating respiratory induced organ motion by vector volume histogram. Phys Med 2016;32(12):1570–4. Wilbert J, Baier K, Hermann C, Flentje M, Guckenberger M. Accuracy of real-time couch tracking during 3-dimensional conformal radiation therapy, intensity modulated radiation therapy, and volumetric modulated arc therapy for prostate cancer. Int J Radiat Oncol Biol Phys 2013;85(1):237–42. Hansen R, Ravkilde T, Worm E, et al. Electromagnetic guided couch and multileaf collimator tracking on a TrueBeam accelerator. Med Phys 2016;43(5):2387–98. Lang S, Zeimetz J, Ochsner G, Schmid Daners M, Riesterer O, Klöck S. Development and evaluation of a prototype tracking system using the treatment couch. Med Phys 2014;41(2):021720. Worm E, Høyer M, Fledelius W, Nielsen J, Larsen L, Poulsen P. On-line use of threedimensional marker trajectory estimation from cone-beam computed tomography projections for precise setup in radiotherapy for targets with respiratory motion. Int J Radiat Oncol Biol Phys 2012;83(1):e145–51. Dinkel J, Hintze C, Tetzlaff R, et al. 4D-MRI analysis of lung tumor motion in patients with hemidiaphragmatic paralysis. Radiother Oncol 2009;91(3):449–54. Boussuges A, Gole Y, Blanc P. Diaphragmatic motion studied by M-mode ultrasonography. Chest 2009;135(2):391–400. Haji K, Royse A, Tharmaraj D, Haji D, Botha J, Royse C. Diaphragmatic regional displacement assessed by ultrasound and correlated to subphrenic organ movement in the critically ill patients-an observational study. J Crit Care 2015;30(2). 439.e7439.e13. Fast M, O'Shea T, Nill S, Oelfke U, Harris E. First evaluation of the feasibility of MLC tracking using ultrasound motion estimation. Med Phys 2016;43(8):4628–33.
18