- Email: [email protected]
Moving targets: detection and tracking of internal organ motion for treatment planning and patient set-up
$68
Moving targets: detection and tracking of internal organ motion for treatment planning and patient set-up Eike Rietzel 1'2, Stanley J. Rosenthai1, David P. Gierga~, Christopher G. Willett~'3, George T.Y. Chen ~ (1) Department of Radiation Ontology, Massachusetts General Hospital, Boston, USA (2) Abteilung Biophysik, Gesellschaft for Schwerionenforschung, Darmstadt, Germany (3) current address: Department of Radiation Oncology, Duke University, Durham, NC Corresponding author Massachusetts General Hospital Northeast Proton Therapy Center Eike Rietzel 30 Fruit Street Boston, MA, 02114 USA Tel.: 617 - 726 - 7895 Fax: 617 - 724 - 0368 e-mail: [email protected]
Summary Background and Purpose Clinical target volumes of the thorax and abdomen are typically expanded to account for inter- and intrafractional organ motion. Usually, such expansions are based on clinical experience and planar observations of target motion during simulation. More precise, 4- dimensional motion margins for a specific patient may improve dose coverage of mobile targets and yet limit unnecessarily large field expansions. We are studying approaches to targeting moving tumors throughout the entire treatment process, from treatment planning to beam delivery. Material and Methods Radio-opaque markers were implanted under CT guidance in the liver at the gross tumor periphery. Organ motion during light respiration was volumetrically imaged by 4D Computed Tomography. Marker metion was also acquired by fluoroscopy and compared with 4DCT data. During treatment, daily diagnostic x-ray images were captured at end-exhale and -inhale for patient setup and target localization. Results Based on the time-resolved CT data, target volumes can be designed to account for respiratory motion during treatment. Motion of the tumor as derived from 4DCT was consistent with fluoroscopic motion analysis. Radiographs acquired in the treatment room enabled millimeter-level patient set-up and assessment of target position relative to bony anatomy. Daily positional variations between bony anatomy and implanted markers were observed. Conclusions Image guided therapy, based on 4DCT imaging, fluoroscopic imaging studies, and daily gated diagonstic energy set-up radiographs is being developed to improve beam delivery precision. Monitoring internal target motion throughout the entire treatment process will ensure adequate dose coverage of the target while sparing the maximum healthy tissue.
Key words organ motion, respiratory motion, 4D Computed Tomography, motion tracking, patient set-up
Introduction Internal organ motion limits the achievable precision in radiotherapy. Organ motion occurs both inter- and intrafractionally. Intrafractional motion is a result of respiration, peristalsis, and cardiac motion. The magnitude of interfractional target motion is dependent on daily variations in organ filling, e.g. rectum or bladder, weight loss or gain, tumor growth or shrinkage, and radiation induced changes of tissue. In current standard practice, the clinical target volume is expanded to account for possible target motion during and between treatment fractions [5,6]. Such expansions are most common tumor site specific based on experience and reported values. Expansions of the clinical target volume result in an increase of the irradiated volume and raise the risk of radiation toxicity to normal tissue. In general, the maximum dose delivered to the target is limited by the tolerance dose of adjacent critical normal organs. Assessing and adapting to patient specific target motion throughout the course of radiotherapy offers the possibility of ensuring delivery of the prescribed target dose while simultaneously minimizing normal tissue damage. Current site-specific target expansions are mainly based on clinical experience. The geometric margin applied to a specific case is based on both population average motion as reported in the literature as well as on estimates of patient specific tumor motion as visualized during simulation. It is well known that internal organ motion parameters show large inter-patient variations [8]. Geometric margins are not typically optimised for each specific patient, and often are isotropic. We have begun to assess the degree of internal target motion patient specifically during planning and simulation process and utilize daily imaging throughout the treatment course. Thus far, this involves time-resolved CT imaging for treatment planning, fluoroscopic tumor motion studies, and daily patient set-up radiographs imaging the internal target. On-line target tracking and improved beam delivery techniques are currently under development [2,10]. In this report we present a case study of a patient with a hepatocellular tumor. 4D Computed Tomography, quantitative fluoroscopic target motion analysis, and daily diagnostic x-ray acquisition for patient positioning will be described. Treatment planning and target volume design are beyond the scope of this limited report but will be addressed elsewhere [13]. This specific patient was treated with proton beam therapy at the Northeast Proton Therapy Center (NPTC) at MGH. 4DCT studies were performed under an investigational protocol approved by the Internal Review Board, fluoroscopy is standard practice to assess internal motion, and acquisition of daily x-ray radiographs is performed routinely at NPTC during patient set-up.
$69
Material and Methods 4D Computed Tomography The patient was scanned in axial cine mode on a 4-slice General Electric (GE) Lightspeed QX/i CT scanner. At each couch position data were acquired continuously during several tube rotations for a time interval equal to the duration of the patient's respiratory cycle. After completion of scanning at the initial couch position, the x-rays were automatically turned off, and the couch was advanced to the next position. This procedure was repeated until the full length of the volume of interest was scanned. At each couch position, multiple images were reconstructed (typically ~20), evenly distributed in time over the respiratory cycle. Each image represents patient anatomy at a specific respiratory phase. 4DCT data acquisition is described in detail in [12]. During CT data acquisition, the patient's abdominal surface motion is recorded by a camera system (Varian, RPM-system). Surface motion and CT data acquisition are precisely correlated in time via an external signal sent from the scanner to the RPM-system. Images are retrospectively sorted into tempo-spatial coherent CT volumes using GE Advantage4D software. The temporal correlation between reconstructed images and surface motion trace is used to assign a specific respiratory phase to each reconstructed image. Specific phases of the respiratory cycle can be selected and the software automatically collects and combines images at or near the specific respiratory phase to form the complete anatomic volume for that phase. The sorting process typically results in 10-20 complete volumetric CT data sets of a patient, each representing a different respiratory phase.
Fluoroscopy Radio-opaque markers were implanted in the liver near the tumor under CT guidance. Fluoroscopic data were acquired during multiple respiratory cycles on a Varian Ximatron on a different day than the 4DCT. Fluoroscopic data described here were recorded with a commercial consumer VCR connected to the image intensifier video signal. During fluoroscopy, additional radio-opaque markers were placed on the patient's abdominal surface to study the correlation between internal and external motion. Results of this analysis will be reported in a separate paper (Gierga, in preparation). Motion data on clip movement during respiration was digitized and analyzed by machine vision clip tracking software [4]. Marker motion is tracked with normalized cross correlation at the full frame rate of 30 Hz.
Daily patient positioning At the NPTC, each patient is initially set up by a laser system based on conventional skin tattoos. Prior to each treatment, orthogonal, diagnostic x-ray images are acquired to perform millimeter-level patient set-up based on bony anatomy. X-rays are captured digitally and compared to digitally reconstructed radiographs (DRR). Landmarks selected on the DRRs are manually marked on the daily x-ray images. Landmarks are categorized to yield axis specific translations or rotations. This assignment or association with movement in a given direction depends on how precise each landmark can contribute to a given motion. For example, the anterior edge of the vertebral column is used primarily to guide AP positional alignment. Inter-vertebral spaces are utilized to determine craniocaudal adjustment. Based on the selected points, and predefinition of how points contribute to the final move, the software determines the position of the iso-center within the patient anatomy and compares it to the room iso-center. Couch translation and rotation are automatically calculated and displayed. Two separate sets of landmarks can be selected resulting in independent iso-center calculations. Within this study, x-rays were acquired during light respiration, either at end-exhale or end-inhale to compare marker positions with DRRs obtained from,4DCT data sets at end-exhale and end-inhale. Acquisition was triggered manually while observing the patient's abdominal surface motion on a video monitor. For our study, bony landmarks were chosen as the primary landmarks. Secondary landmarks were the radio-opaque markers in the liver to analyse their positions in relation to bony anatomy. Pre-treatment patient alignment was performed according to the NPTC standard protocol using bony landmarks. Marker positions were analysed only.
Results 4D Computed Tomography Figure 1 shows three different respiratory phases at the same coronal plane obtained through 4DCT imaging, A horizontal line is overlaid to guide the eye in assessing organ motion. Craniocaudal peak-to-peak motion at the top of the liver was measured t o be about 25 mm. One of the four radio-opaque markers in the liver is visible in this slice (marker 2, see figure 3a). The respiratory phase closest to inhale (left) shows minor residual artifacts at the superior margin of the liver; these artefacts are a result of irregular respiration during the 4DCT data acquisition. Note the absence of typical, well-known motion artifacts such as significant distortions at the top of the liver compared to a standard CT scan acquired during light respiration. Figure 1
Figure 1: Internal organ motion imaged with 4D Computed Tomography, same coronal slice, different volumes. Images correspond from left to right to inhale, mid-exhale, and exhale. A horizontal line is plotted to guide the eye.
$70
Radio.opaque marker trajectories Marker positions during the respiratory cycle were extracted in 3D from 10 4DCT volumes. This is done by examining the 3D coordinates of the specific clip in each of the 10 CT data sets generated by 4DCT. The trajectory over a respiratory cycle is shown in figure 2 (bottom). Marker positions are shown in figure 3a on a DRR. Coordinates extracted from Fluoroscopic data of marker 4 (lateral view) are displayed in figure 2 (top). Other markers could not be tracked in the fluoroscopic lateral view, because their visibility could not reliably identified due to thicker parts of the patient. Marker motion is summarized in table 1. The 4DCT trajectories for marker 4 are consistent with the excursions of clip 4 obtained from fluoroscopic analysis. The 4DCT trajectories of the markers show reasonable motion patterns. All markers move inferiorly and anteriorly at inhale. Lateral motion differs for different markers, depending on their position. The superior markers 1 and 2, move to the right while inferior markers 3 and 4 move to the left during inhalation. The overall 3D motion pattern from exhale to inhale appears to be in inferior-anterior direction with a superimposed rotation about an axis in anterior-posterior direction. It should be noted, that this analysis does not include possible deformation of the liver during respiration. However, the motion pattern seems reasonable. Liver motion is affected by the internal patient contour, which is dominated by the shape of the ribcage. Figure 1 shows ribcage curvature resulting in the observed rotational motion. Fi.qure 2
tFigure 2: Trajectories of radio-opaque markers implanted in the liver. Top: lateral fluoroscopic data for one of the markers acquired over approximately five respiratory cycles. Bottom: marker motion images with 4D Computed Tomography for four markers, left to right correspond to superior-inferior, anterior-posterior, and left-right marker motion. Note the different scales. Table 1: Peak-to-peak maximum of radio-opaque marker motions within a respiratory cycle obtained from 4DCT data analysis and maximum and minimum peak-to-peak extent obtained from fluoroscopic marker tracking over approximately 5 respiratory cycles, all values in mm. 4D Computed Tomography
fluoroscopy
SImax
SI
AP
LR
marker 1
22.5
7.3
7.4
.
marker 2
17.6
4.3
2.9
--
marker 3
17.6
8.0
2.9
--
marker 4
10.0
3.6
5.1
13.6
APmax .
.
5.5
Slmin
APrnin
7.2
3.1
.
$71
Daily patient positioning Figure 3 displays a digitally reconstructed radiograph (a) and 3 x-ray radiographs (b-d) acquired during the process of daily image guided patient positioning. The outline of the aperture and the treatment iso-center are displayed in blue. Outlines indicating the radio-opaque markers at end-exhale position are shown in yellow. On these x-ray images, axes for the iso-center based on bony anatomy are displayed in red, according to radio-opaque marker positions in yellow. Fi.qure 3
Figure 3: Daily patient set-up, anterior-posterior view. a) digitally reconstructed radiograph, overlaid aperture and contours of radioopaque markers at exhale, b-d) digitally acquired x-rays for patient set-up; coordinate systems displayed: iso-center in treatment room (blue), iso-center according to bony anatomy (red), and iso-center according to radio-opaque marker positions (yellow). Figure 3b shows an x-ray image acquired at end-inhale during light respiration. According to bony landmarks the couch was moved left (3 mm) and infefiorly (9 mm). Compared to bony anatomy, the iso-center according to radio-opaque markers is offset inferiorly by 22 mm and laterally by 4 mm. Parameters for marker based iso-center determination were set according to marker positions at end-exhale. The differences between bony anatomy and marker positions at end-inhale are in good agreement with the amplitudes of respiratory marker motion obtained by 4DCT and fluoroscopy. Figure 3c shows an x-ray image acquired at end-exhale during light respiration. Iso-centers determined from bony landmarks and from marker positions are therefore expected to be in good agreement. However, differences are 9 mm in inferior and 3 mm in lateral direction. Such differences occur most likely due to more shallow respiration compared to imaging or due to wrong timing for the x-ray acquisition. The observed difference should not pose a problem, marker positions are within the range of motion imaged with 4DCT. The x-ray image shown in figure 3d was acquired at end-exhale on a different day. Bony landmark and marker-based iso-centers agree in craniocaudal direction while in the lateral direction they differ by 4 mm,
Discussion Internal target location of lesions in the abdomen and thorax is subject to inter- and intrafractional variations. To design patient specific target volumes, motion must be controlled or monitored throughout the entire treatment process. The most precise motion management method to simultaneously account for both inter- and intrafractional target motion during treatment is fluoroscopic marker tracking [14]. However, the required hardware and software for this is currently not commonly available. We have taken a simpler approach using commercially available hardware to treat moving targets. In this process, imaging is used throughout the steps, from the planning scan to treatment. 4D Computed Tomography was used to image respiration induced internal motion in 3D as a function of time. The advantage of this imaging technique over conventional fluoroscopic studies is that 1)
S72
4DCT enables the physician to appreciate the entire range of motion of the GTV / clips in 4D 2) 4DCT provides information on the motion of critical structures not visible in fluoroscopy (not clipped) 3) fluoroscopic video usually images a single projection at a time (orthogonal simultaneous fluoroscopy is not generally available). The advantage of fluoroscopic video is that motion can be measured for many breath cycles, and permits the assessment of internal motion variation during normal breathing. Furthermore the correlation between internal target and external surface motion can be assessed which will be useful for treatments that are gated based on surface motion. Comparisons of fluoroscopic marker tracking and 4DCT are consistent, suggesting reasonable evidence for consistent respiration, considering the data were acquired on different days. Fluoroscopic imaging data in this case study show that tumor motion amplitudes can significantly differ even within the same fraction, e.g. for the patient in this case study by approximately 6 mm craniocaudally. Motion derived from 4DCT data is comparable to maximum amplitudes imaged with fluoroscopy. Target delineation based on 4DCT data therefore should encompass most of the intrafractional motion variations. Using radio-opaque markers, interfractional target motion can be adjusted for by daily diagnostic x-ray imaging at a well defined respiratory phase. In this study, x-rays were acquired by manual trigger while observing the abdominal surface, a surrogate for respiratory phase. In future, the radiographs will be acquired more accurately by automatically triggering based on a respiratory signal. Monitoring the respiratory pattern to assess its regularity will ensure that x-ray images are not acquired during an unusual instance such as during shallow or deep breath, cough etc. 4DCT data can be utilized to determine the appropriate treatment technique and target expansion depending on the magnitude of motion observed. For limited respiratory target motion, a simple expansion strategy seems reasonable. The dosimetric gains of gated [11] or breath-controlled [9,15] treatments can be directly analysed based on time-resolved volumetric data. Specifically, 4D dose distributions for a given gating duty-cycle can be calculated by combining target volumes for respiratory phases within the gating window. It is essential to validate that respiratory induced target motion as imaged by 4DCT is an accurate representation of motion throughout the treatment. Internal target position can be imaged in the treatment room on a daily basis by a variety of other imaging systems as well, e.g. in-room CT scanners [3] or cone-beam CT [7]. For targets influenced by respiration, acquisition of xray radiographs seems to be advantageous compared to in-room CT or cone-beam acquisitions - although in principle timeresolved data acquisition with such systems is possible as well. Radiographs can easily be acquired at pre-selected respiratory phases whereas conventional CT scanning will be subject to motion artifacts. For the case study presented here, daily internal target positions were within the margins derived from 4DCT. Using x-ray imaging without radio-opaque markers, daily internal target position can be estimated from a combination of diaphragm position and bony anatomy at selected respiratory phases or at suspended respiration [1]. During treatment deliveries respiratory motion should be recorded to validate compliance with the anatomic model used for treatment planning. Suitable interaction levels for deviations found for interfractional marker positions and respiratory variations outside the patient specific margins used for treatment planning will have to be established. Furthermore, strategies to deal with possible deviations will have to be developed. References [1] Baiter JM, Brock KK, Litzenberg DW, McShan DE, Lawrence TS, Ten Haken R, McGinn CJ, Lam KL, Dawson LA. Daily targeting of intrahepatic tumors for radiotherapy. Int J Radiat Oncol Biol Phys 2002;52:266-71. [2] Berbeco R, Jiang SB, Sharp GC, Chen GTY, Mostafavi H, Shirato H. Integrated radiotherapy imaging system (IRIS): design considerations of tumor tracking with linac gantry-mounted diagnostic x-ray systems with flat-panel detectors. Phys Med Biol 2004;49:243-55. [3] Cheng CW, Wong J, Grimm L, Chow M, Uematsu M, Fung A. Commissioning and clinical implementation of a sliding gantry CT scanner installed in an existing treatment room and early clinical experience for precise tumor localization. Am J Clin Qncol 2003;26:e28-36. [4] Gierga DP, Chert GTY, Kung JH, Betke M, Lombardi J, Willett CG. Quantification of respiration-induced abdominal tumor motion and its impact on IMRT dose distributions. Int J Radiat Oncol Biol Phys 2004:58;1584-1595. [5] ICRU50. Prescribing, Recording and Reporting Photon Beam Therapy. ICRU, Bethesda, MD, 1993. [6] ICRU62. Prescribing, Recording and Reporting Photon Beam Therapy (supplement to ICRU Report 50). ICRU, Bethesda, MD, 1999. [7] Jaffray DA, Siewerdsen JH, Wong JW, Martinez AA. Flat-panel cone-beam computed tomography for image-guided radiation therapy. Int J Radiat Oncol Biol Phys 2002;53:1337-49. [8] Langen KM, Jones DT. Organ motion and its management. Int J Radiat Oncol Biol Phys 2001 ;50:265-78.
[9] Margeras GS, Yorke E. Deep inspiration breath hold and respiratory gating strategies for reducing organ motion in radiation treatment. Semin Radiat Qnco12004;14:65-75. [10] Neicu T, Shirato H, Seppenwoolde Y, Jiang SB. Synchronized moving aperture radiation therapy (SMART): average tumour trajectory for lung patients. Phys Med Bio12003;48:587-98. [1 I] Ohara K, Okumura T, Akisada M. Irradiation synchronized with respiration gate. Int J Radiat Oncol Biol Phys 1989;17:853-857. [12] Pan T, Lee TY, Rietzel E, Chen GT. 4D-CT imaging of a volume influenced by respiratory motion on multi-slice CT. Med Phys 2004;31:333-40. [13] Rietzel E, Chen GTY, Choi NC, Willett CG. 4D Image Based Treatment Planning: Target Volume Segmentation and Dose Calculation in the Presence of Respiratory Motion. Int J Radiat Oncol Biol Phys, submitted. [14] Shirato H, Shimizu S, Kitamura K, Nishioka T, Kagei K, Hashimoto S, Aoyama H, Kunieda T, Shinohara N, Dosaka-Akita H, Miyasaka K. Four-dimensional treatment planning and fluoroscopic real-time tumor tracking radiotherapy for moving tumor. Int J Radiat Oncol Biol Phys 2000;48:435-42. [15] Wong JW, Sharpe MB, Jaffray DA, Kini VR, Robertson JM, Stromberg JS, Martinez AA. The use of active breathing control (ABC) to reduce margin for breathing motion. Int J Radiat Oncol Biol Phys 1999;44:911-9.
Moving targets: detection and tracking of internal organ motion for treatment planning and patient set-up Eike Rietzel 1'2, Stanley J. Rosenthai1, David P. Gierga~, Christopher G. Willett~'3, George T.Y. Chen ~ (1) Department of Radiation Ontology, Massachusetts General Hospital, Boston, USA (2) Abteilung Biophysik, Gesellschaft for Schwerionenforschung, Darmstadt, Germany (3) current address: Department of Radiation Oncology, Duke University, Durham, NC Corresponding author Massachusetts General Hospital Northeast Proton Therapy Center Eike Rietzel 30 Fruit Street Boston, MA, 02114 USA Tel.: 617 - 726 - 7895 Fax: 617 - 724 - 0368 e-mail: [email protected]
Summary Background and Purpose Clinical target volumes of the thorax and abdomen are typically expanded to account for inter- and intrafractional organ motion. Usually, such expansions are based on clinical experience and planar observations of target motion during simulation. More precise, 4- dimensional motion margins for a specific patient may improve dose coverage of mobile targets and yet limit unnecessarily large field expansions. We are studying approaches to targeting moving tumors throughout the entire treatment process, from treatment planning to beam delivery. Material and Methods Radio-opaque markers were implanted under CT guidance in the liver at the gross tumor periphery. Organ motion during light respiration was volumetrically imaged by 4D Computed Tomography. Marker metion was also acquired by fluoroscopy and compared with 4DCT data. During treatment, daily diagnostic x-ray images were captured at end-exhale and -inhale for patient setup and target localization. Results Based on the time-resolved CT data, target volumes can be designed to account for respiratory motion during treatment. Motion of the tumor as derived from 4DCT was consistent with fluoroscopic motion analysis. Radiographs acquired in the treatment room enabled millimeter-level patient set-up and assessment of target position relative to bony anatomy. Daily positional variations between bony anatomy and implanted markers were observed. Conclusions Image guided therapy, based on 4DCT imaging, fluoroscopic imaging studies, and daily gated diagonstic energy set-up radiographs is being developed to improve beam delivery precision. Monitoring internal target motion throughout the entire treatment process will ensure adequate dose coverage of the target while sparing the maximum healthy tissue.
Key words organ motion, respiratory motion, 4D Computed Tomography, motion tracking, patient set-up
Introduction Internal organ motion limits the achievable precision in radiotherapy. Organ motion occurs both inter- and intrafractionally. Intrafractional motion is a result of respiration, peristalsis, and cardiac motion. The magnitude of interfractional target motion is dependent on daily variations in organ filling, e.g. rectum or bladder, weight loss or gain, tumor growth or shrinkage, and radiation induced changes of tissue. In current standard practice, the clinical target volume is expanded to account for possible target motion during and between treatment fractions [5,6]. Such expansions are most common tumor site specific based on experience and reported values. Expansions of the clinical target volume result in an increase of the irradiated volume and raise the risk of radiation toxicity to normal tissue. In general, the maximum dose delivered to the target is limited by the tolerance dose of adjacent critical normal organs. Assessing and adapting to patient specific target motion throughout the course of radiotherapy offers the possibility of ensuring delivery of the prescribed target dose while simultaneously minimizing normal tissue damage. Current site-specific target expansions are mainly based on clinical experience. The geometric margin applied to a specific case is based on both population average motion as reported in the literature as well as on estimates of patient specific tumor motion as visualized during simulation. It is well known that internal organ motion parameters show large inter-patient variations [8]. Geometric margins are not typically optimised for each specific patient, and often are isotropic. We have begun to assess the degree of internal target motion patient specifically during planning and simulation process and utilize daily imaging throughout the treatment course. Thus far, this involves time-resolved CT imaging for treatment planning, fluoroscopic tumor motion studies, and daily patient set-up radiographs imaging the internal target. On-line target tracking and improved beam delivery techniques are currently under development [2,10]. In this report we present a case study of a patient with a hepatocellular tumor. 4D Computed Tomography, quantitative fluoroscopic target motion analysis, and daily diagnostic x-ray acquisition for patient positioning will be described. Treatment planning and target volume design are beyond the scope of this limited report but will be addressed elsewhere [13]. This specific patient was treated with proton beam therapy at the Northeast Proton Therapy Center (NPTC) at MGH. 4DCT studies were performed under an investigational protocol approved by the Internal Review Board, fluoroscopy is standard practice to assess internal motion, and acquisition of daily x-ray radiographs is performed routinely at NPTC during patient set-up.
$69
Material and Methods 4D Computed Tomography The patient was scanned in axial cine mode on a 4-slice General Electric (GE) Lightspeed QX/i CT scanner. At each couch position data were acquired continuously during several tube rotations for a time interval equal to the duration of the patient's respiratory cycle. After completion of scanning at the initial couch position, the x-rays were automatically turned off, and the couch was advanced to the next position. This procedure was repeated until the full length of the volume of interest was scanned. At each couch position, multiple images were reconstructed (typically ~20), evenly distributed in time over the respiratory cycle. Each image represents patient anatomy at a specific respiratory phase. 4DCT data acquisition is described in detail in [12]. During CT data acquisition, the patient's abdominal surface motion is recorded by a camera system (Varian, RPM-system). Surface motion and CT data acquisition are precisely correlated in time via an external signal sent from the scanner to the RPM-system. Images are retrospectively sorted into tempo-spatial coherent CT volumes using GE Advantage4D software. The temporal correlation between reconstructed images and surface motion trace is used to assign a specific respiratory phase to each reconstructed image. Specific phases of the respiratory cycle can be selected and the software automatically collects and combines images at or near the specific respiratory phase to form the complete anatomic volume for that phase. The sorting process typically results in 10-20 complete volumetric CT data sets of a patient, each representing a different respiratory phase.
Fluoroscopy Radio-opaque markers were implanted in the liver near the tumor under CT guidance. Fluoroscopic data were acquired during multiple respiratory cycles on a Varian Ximatron on a different day than the 4DCT. Fluoroscopic data described here were recorded with a commercial consumer VCR connected to the image intensifier video signal. During fluoroscopy, additional radio-opaque markers were placed on the patient's abdominal surface to study the correlation between internal and external motion. Results of this analysis will be reported in a separate paper (Gierga, in preparation). Motion data on clip movement during respiration was digitized and analyzed by machine vision clip tracking software [4]. Marker motion is tracked with normalized cross correlation at the full frame rate of 30 Hz.
Daily patient positioning At the NPTC, each patient is initially set up by a laser system based on conventional skin tattoos. Prior to each treatment, orthogonal, diagnostic x-ray images are acquired to perform millimeter-level patient set-up based on bony anatomy. X-rays are captured digitally and compared to digitally reconstructed radiographs (DRR). Landmarks selected on the DRRs are manually marked on the daily x-ray images. Landmarks are categorized to yield axis specific translations or rotations. This assignment or association with movement in a given direction depends on how precise each landmark can contribute to a given motion. For example, the anterior edge of the vertebral column is used primarily to guide AP positional alignment. Inter-vertebral spaces are utilized to determine craniocaudal adjustment. Based on the selected points, and predefinition of how points contribute to the final move, the software determines the position of the iso-center within the patient anatomy and compares it to the room iso-center. Couch translation and rotation are automatically calculated and displayed. Two separate sets of landmarks can be selected resulting in independent iso-center calculations. Within this study, x-rays were acquired during light respiration, either at end-exhale or end-inhale to compare marker positions with DRRs obtained from,4DCT data sets at end-exhale and end-inhale. Acquisition was triggered manually while observing the patient's abdominal surface motion on a video monitor. For our study, bony landmarks were chosen as the primary landmarks. Secondary landmarks were the radio-opaque markers in the liver to analyse their positions in relation to bony anatomy. Pre-treatment patient alignment was performed according to the NPTC standard protocol using bony landmarks. Marker positions were analysed only.
Results 4D Computed Tomography Figure 1 shows three different respiratory phases at the same coronal plane obtained through 4DCT imaging, A horizontal line is overlaid to guide the eye in assessing organ motion. Craniocaudal peak-to-peak motion at the top of the liver was measured t o be about 25 mm. One of the four radio-opaque markers in the liver is visible in this slice (marker 2, see figure 3a). The respiratory phase closest to inhale (left) shows minor residual artifacts at the superior margin of the liver; these artefacts are a result of irregular respiration during the 4DCT data acquisition. Note the absence of typical, well-known motion artifacts such as significant distortions at the top of the liver compared to a standard CT scan acquired during light respiration. Figure 1
Figure 1: Internal organ motion imaged with 4D Computed Tomography, same coronal slice, different volumes. Images correspond from left to right to inhale, mid-exhale, and exhale. A horizontal line is plotted to guide the eye.
$70
Radio.opaque marker trajectories Marker positions during the respiratory cycle were extracted in 3D from 10 4DCT volumes. This is done by examining the 3D coordinates of the specific clip in each of the 10 CT data sets generated by 4DCT. The trajectory over a respiratory cycle is shown in figure 2 (bottom). Marker positions are shown in figure 3a on a DRR. Coordinates extracted from Fluoroscopic data of marker 4 (lateral view) are displayed in figure 2 (top). Other markers could not be tracked in the fluoroscopic lateral view, because their visibility could not reliably identified due to thicker parts of the patient. Marker motion is summarized in table 1. The 4DCT trajectories for marker 4 are consistent with the excursions of clip 4 obtained from fluoroscopic analysis. The 4DCT trajectories of the markers show reasonable motion patterns. All markers move inferiorly and anteriorly at inhale. Lateral motion differs for different markers, depending on their position. The superior markers 1 and 2, move to the right while inferior markers 3 and 4 move to the left during inhalation. The overall 3D motion pattern from exhale to inhale appears to be in inferior-anterior direction with a superimposed rotation about an axis in anterior-posterior direction. It should be noted, that this analysis does not include possible deformation of the liver during respiration. However, the motion pattern seems reasonable. Liver motion is affected by the internal patient contour, which is dominated by the shape of the ribcage. Figure 1 shows ribcage curvature resulting in the observed rotational motion. Fi.qure 2
tFigure 2: Trajectories of radio-opaque markers implanted in the liver. Top: lateral fluoroscopic data for one of the markers acquired over approximately five respiratory cycles. Bottom: marker motion images with 4D Computed Tomography for four markers, left to right correspond to superior-inferior, anterior-posterior, and left-right marker motion. Note the different scales. Table 1: Peak-to-peak maximum of radio-opaque marker motions within a respiratory cycle obtained from 4DCT data analysis and maximum and minimum peak-to-peak extent obtained from fluoroscopic marker tracking over approximately 5 respiratory cycles, all values in mm. 4D Computed Tomography
fluoroscopy
SImax
SI
AP
LR
marker 1
22.5
7.3
7.4
.
marker 2
17.6
4.3
2.9
--
marker 3
17.6
8.0
2.9
--
marker 4
10.0
3.6
5.1
13.6
APmax .
.
5.5
Slmin
APrnin
7.2
3.1
.
$71
Daily patient positioning Figure 3 displays a digitally reconstructed radiograph (a) and 3 x-ray radiographs (b-d) acquired during the process of daily image guided patient positioning. The outline of the aperture and the treatment iso-center are displayed in blue. Outlines indicating the radio-opaque markers at end-exhale position are shown in yellow. On these x-ray images, axes for the iso-center based on bony anatomy are displayed in red, according to radio-opaque marker positions in yellow. Fi.qure 3
Figure 3: Daily patient set-up, anterior-posterior view. a) digitally reconstructed radiograph, overlaid aperture and contours of radioopaque markers at exhale, b-d) digitally acquired x-rays for patient set-up; coordinate systems displayed: iso-center in treatment room (blue), iso-center according to bony anatomy (red), and iso-center according to radio-opaque marker positions (yellow). Figure 3b shows an x-ray image acquired at end-inhale during light respiration. According to bony landmarks the couch was moved left (3 mm) and infefiorly (9 mm). Compared to bony anatomy, the iso-center according to radio-opaque markers is offset inferiorly by 22 mm and laterally by 4 mm. Parameters for marker based iso-center determination were set according to marker positions at end-exhale. The differences between bony anatomy and marker positions at end-inhale are in good agreement with the amplitudes of respiratory marker motion obtained by 4DCT and fluoroscopy. Figure 3c shows an x-ray image acquired at end-exhale during light respiration. Iso-centers determined from bony landmarks and from marker positions are therefore expected to be in good agreement. However, differences are 9 mm in inferior and 3 mm in lateral direction. Such differences occur most likely due to more shallow respiration compared to imaging or due to wrong timing for the x-ray acquisition. The observed difference should not pose a problem, marker positions are within the range of motion imaged with 4DCT. The x-ray image shown in figure 3d was acquired at end-exhale on a different day. Bony landmark and marker-based iso-centers agree in craniocaudal direction while in the lateral direction they differ by 4 mm,
Discussion Internal target location of lesions in the abdomen and thorax is subject to inter- and intrafractional variations. To design patient specific target volumes, motion must be controlled or monitored throughout the entire treatment process. The most precise motion management method to simultaneously account for both inter- and intrafractional target motion during treatment is fluoroscopic marker tracking [14]. However, the required hardware and software for this is currently not commonly available. We have taken a simpler approach using commercially available hardware to treat moving targets. In this process, imaging is used throughout the steps, from the planning scan to treatment. 4D Computed Tomography was used to image respiration induced internal motion in 3D as a function of time. The advantage of this imaging technique over conventional fluoroscopic studies is that 1)
S72
4DCT enables the physician to appreciate the entire range of motion of the GTV / clips in 4D 2) 4DCT provides information on the motion of critical structures not visible in fluoroscopy (not clipped) 3) fluoroscopic video usually images a single projection at a time (orthogonal simultaneous fluoroscopy is not generally available). The advantage of fluoroscopic video is that motion can be measured for many breath cycles, and permits the assessment of internal motion variation during normal breathing. Furthermore the correlation between internal target and external surface motion can be assessed which will be useful for treatments that are gated based on surface motion. Comparisons of fluoroscopic marker tracking and 4DCT are consistent, suggesting reasonable evidence for consistent respiration, considering the data were acquired on different days. Fluoroscopic imaging data in this case study show that tumor motion amplitudes can significantly differ even within the same fraction, e.g. for the patient in this case study by approximately 6 mm craniocaudally. Motion derived from 4DCT data is comparable to maximum amplitudes imaged with fluoroscopy. Target delineation based on 4DCT data therefore should encompass most of the intrafractional motion variations. Using radio-opaque markers, interfractional target motion can be adjusted for by daily diagnostic x-ray imaging at a well defined respiratory phase. In this study, x-rays were acquired by manual trigger while observing the abdominal surface, a surrogate for respiratory phase. In future, the radiographs will be acquired more accurately by automatically triggering based on a respiratory signal. Monitoring the respiratory pattern to assess its regularity will ensure that x-ray images are not acquired during an unusual instance such as during shallow or deep breath, cough etc. 4DCT data can be utilized to determine the appropriate treatment technique and target expansion depending on the magnitude of motion observed. For limited respiratory target motion, a simple expansion strategy seems reasonable. The dosimetric gains of gated [11] or breath-controlled [9,15] treatments can be directly analysed based on time-resolved volumetric data. Specifically, 4D dose distributions for a given gating duty-cycle can be calculated by combining target volumes for respiratory phases within the gating window. It is essential to validate that respiratory induced target motion as imaged by 4DCT is an accurate representation of motion throughout the treatment. Internal target position can be imaged in the treatment room on a daily basis by a variety of other imaging systems as well, e.g. in-room CT scanners [3] or cone-beam CT [7]. For targets influenced by respiration, acquisition of xray radiographs seems to be advantageous compared to in-room CT or cone-beam acquisitions - although in principle timeresolved data acquisition with such systems is possible as well. Radiographs can easily be acquired at pre-selected respiratory phases whereas conventional CT scanning will be subject to motion artifacts. For the case study presented here, daily internal target positions were within the margins derived from 4DCT. Using x-ray imaging without radio-opaque markers, daily internal target position can be estimated from a combination of diaphragm position and bony anatomy at selected respiratory phases or at suspended respiration [1]. During treatment deliveries respiratory motion should be recorded to validate compliance with the anatomic model used for treatment planning. Suitable interaction levels for deviations found for interfractional marker positions and respiratory variations outside the patient specific margins used for treatment planning will have to be established. Furthermore, strategies to deal with possible deviations will have to be developed. References [1] Baiter JM, Brock KK, Litzenberg DW, McShan DE, Lawrence TS, Ten Haken R, McGinn CJ, Lam KL, Dawson LA. Daily targeting of intrahepatic tumors for radiotherapy. Int J Radiat Oncol Biol Phys 2002;52:266-71. [2] Berbeco R, Jiang SB, Sharp GC, Chen GTY, Mostafavi H, Shirato H. Integrated radiotherapy imaging system (IRIS): design considerations of tumor tracking with linac gantry-mounted diagnostic x-ray systems with flat-panel detectors. Phys Med Biol 2004;49:243-55. [3] Cheng CW, Wong J, Grimm L, Chow M, Uematsu M, Fung A. Commissioning and clinical implementation of a sliding gantry CT scanner installed in an existing treatment room and early clinical experience for precise tumor localization. Am J Clin Qncol 2003;26:e28-36. [4] Gierga DP, Chert GTY, Kung JH, Betke M, Lombardi J, Willett CG. Quantification of respiration-induced abdominal tumor motion and its impact on IMRT dose distributions. Int J Radiat Oncol Biol Phys 2004:58;1584-1595. [5] ICRU50. Prescribing, Recording and Reporting Photon Beam Therapy. ICRU, Bethesda, MD, 1993. [6] ICRU62. Prescribing, Recording and Reporting Photon Beam Therapy (supplement to ICRU Report 50). ICRU, Bethesda, MD, 1999. [7] Jaffray DA, Siewerdsen JH, Wong JW, Martinez AA. Flat-panel cone-beam computed tomography for image-guided radiation therapy. Int J Radiat Oncol Biol Phys 2002;53:1337-49. [8] Langen KM, Jones DT. Organ motion and its management. Int J Radiat Oncol Biol Phys 2001 ;50:265-78.
[9] Margeras GS, Yorke E. Deep inspiration breath hold and respiratory gating strategies for reducing organ motion in radiation treatment. Semin Radiat Qnco12004;14:65-75. [10] Neicu T, Shirato H, Seppenwoolde Y, Jiang SB. Synchronized moving aperture radiation therapy (SMART): average tumour trajectory for lung patients. Phys Med Bio12003;48:587-98. [1 I] Ohara K, Okumura T, Akisada M. Irradiation synchronized with respiration gate. Int J Radiat Oncol Biol Phys 1989;17:853-857. [12] Pan T, Lee TY, Rietzel E, Chen GT. 4D-CT imaging of a volume influenced by respiratory motion on multi-slice CT. Med Phys 2004;31:333-40. [13] Rietzel E, Chen GTY, Choi NC, Willett CG. 4D Image Based Treatment Planning: Target Volume Segmentation and Dose Calculation in the Presence of Respiratory Motion. Int J Radiat Oncol Biol Phys, submitted. [14] Shirato H, Shimizu S, Kitamura K, Nishioka T, Kagei K, Hashimoto S, Aoyama H, Kunieda T, Shinohara N, Dosaka-Akita H, Miyasaka K. Four-dimensional treatment planning and fluoroscopic real-time tumor tracking radiotherapy for moving tumor. Int J Radiat Oncol Biol Phys 2000;48:435-42. [15] Wong JW, Sharpe MB, Jaffray DA, Kini VR, Robertson JM, Stromberg JS, Martinez AA. The use of active breathing control (ABC) to reduce margin for breathing motion. Int J Radiat Oncol Biol Phys 1999;44:911-9.