Relation of external surface to internal tumor motion studied with cine CT

Transcription

1 Relation of external surface to internal tumor motion studied with cine CT Pai-Chun Melinda Chi, a Peter Balter, Dershan Luo, Radhe Mohan, and Tinsu Pan Departments of Imaging Physics and Radiation Physics, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, Texas Received 23 January 2006; revised 30 June 2006; accepted for publication 6 July 2006; published 16 August 2006 The accuracy of delivering gated-radiation therapy to lung tumors using an external respiratory surrogate relies on not only interfractional and intrafractional reproducibility, but also a strong correlation between external motion and internal tumor motion. The purpose of this work was to use the cine images acquired by four-dimensional computed tomography acquisition protocol to study the relation between external surface motion and internal tumor motion. The respiratory phase information of tumor motion and chest wall motion was measured on the cine images using a proposed region-of-interest ROI method and compared to measurement of an external respiratory monitoring device. On eight lung patient data sets, the phase shifts were measured between 1 the signal of a real-time positioning-management RPM respiratory monitoring device placed in the abdominal region and four surface locations on the chest wall, 2 the RPM signal in the abdominal region and tumor motions, and 3 chest wall surface motions and tumor motions. Respiratory waveforms measured at different surface locations during the same respiratory cycle often varied and had significant phase shifts. Seven of the 8 patients showed the abdominal motion leading chest wall motion. The best correlation smallest phase shift was found between the abdominal motion and the superior-inferior S-I tumor motion. A wide range of phase shifts was observed between external surface motion and tumor anterior-posterior A-P /lateral motion. The result supported the placement of the RPM block in the abdominal region and suggested that during a gated therapy utilizing the RPM system, it is necessary to place the RPM block at the same location as it is during treatment simulation in order to reduce potential errors introduced by the position of the RPM block. Correlations between external motions and lateral/a-p tumor motions were inconclusive due to a combination of patient selection and the limitation of the ROI method American Association of Physicists in Medicine. DOI: / Key words: four-dimensional computed tomography, tumor motion, phase shift, respiratory motion I. INTRODUCTION Gated-radiation therapy for lung tumors can provide better tumor targeting and more tissue sparing. 1 One approach to delivering a gated therapy is to use the linear accelerator based system gated by an external respiratory monitor. For such a delivery system, besides the reproducibility of both interfractional and intrafractional motions, 2,3 the treatment precision and accuracy also rely heavily on displacement and temporal correlations between the respiration monitor and tumor motions. 4 Fluoroscopy with 5,6 or without 7,8 the use of fiducial implants is the most commonly employed imaging modality for the motion correlation study. Four-dimensional computed tomography 4DCT has been demonstrated to be useful in allowing the visualization of tumor motion caused by respiration. 9 Underberg et al. have shown that 4DCT accounts for the respiratorycorrelated mobility better than the average of six regular fast CT scans. 10 More recently, 4DCT has been used to characterize the amount of tumor motion 11 and has been incorporated into the radiotherapy treatment planning. 12,13 One can generate the raw 4DCT projection data by employing either the cine 14,15 or helical 16 acquisition protocol. By sampling CT projection data at each slice location for approximately the duration of one respiratory cycle, a set of CT images can be reconstructed for each slice location representing anatomy at different respiratory states. These reconstructed CT images are time stamped and can be correlated to a phase in the breathing cycle monitored by an external device. Through a sorting process, either retrospective 17 or prospective, 18 CT images of the same phase at different axial slices are grouped together to obtain respiratory-correlated CT or 4DCT. In our institution, 4DCT data are acquired using the cine approach, and then retrospectively binned into 10 phases using respiratory phases as detected by real-time positioningmanagement RPM. To form complete volumes in 4DCT, tolerances in respiratory phase are necessary during the binning process because image data are not available for exactly the same phase at each couch position. Resulting volumes are then labeled with a specific phase although they consist of slices that were initially labeled with phases that vary by up to 10% typically 5% 7%. In the current work, we propose to use the cine CT data for studying the temporal relation between external skin motion and tumor motion. The cine CT data are the 4DCT data without the binning process such that they contain no phase error and are therefore more accurate. The purpose is to answer two questions: Is the chest wall a better location than the abdomen for the place Med. Phys. 33 9, September /2006/33 9 /3116/8/$ Am. Assoc. Phys. Med. 3116

2 3117 Chi et al.: External and internal motions 3117 FIG. 1. Schematics of image reconstruction for a slice location. In 4DCT cine acquisition protocol, CT images are reconstructed from 0.5 s of projection data at a small time interval. As shown in the diagram, for a 4-s cine duration with a 0.3-s time interval, there are a total of 12 CT images. ment of an external respiration monitoring device because it is closer to the tumor location, and is there a phase correlation between the external skin motion and tumor motion? II. METHODS AND MATERIALS The study was conducted under the institutional retrospective chart review protocol RCR A. Patient data acquisition Eight patients were selected based on the criteria of having regularly shaped, small to medium solid lung tumors with clear boundaries. These patients were imaged with our institution s 4DCT protocol 15 on a multislice CT scanner Discovery ST; GEHealth Care, Waukesha, WI. The patients were scanned at 0.5 s/ revolution gantry rotation speed, and 2.5-mm slice thickness for eight slices per couch position at 120 kv and ma. The duration of cine acquisition cine duration was the estimated average of the patient s breathing cycle duration plus 1 s. At each couch position, the coverage is 2 cm in the cranio-caudal direction 2.5-mm slice thickness 8 slices and the translational time between couch positions during which x rays were turned off is approximately 1.3 s. Respiratory trace was recorded using the real-time position management block of the RPM system Varian Oncology Systems, Palo Alto, CA as the external respiratory monitoring device. The RPM block was placed in the abdominal region where the amplitude of the respiratory waveform was high. 1. Image reconstruction for 4DCT cine protocol As shown in Fig. 1, the CT projection data for each slice location were acquired continuously over the set cine duration and each CT image was reconstructed from 360 projection data taken over 0.5 s. The reconstructed CT image was assigned the midscan time as the time stamp. The first CT image of the slice labeled IM1 was time stamped the initial time of x ray on plus 0.25 s and the following images were reconstructed at a preset time interval, overlapping some projection data of the previous image. The shorter the image reconstruction interval, the more overlapping projection data is used to reconstruct the CT images. At the time of this study, the reconstruction engine was limited to a total number of CT images per scan series to a maximum of 1500 images. The total number of CT images is the product of the number of slices per couch position, number of CT images per slice, and number of couch positions to cover the length of the scanning volume. This limitation affected how small the image reconstruction interval could be, and further discussion of the uncertainty in measurement caused by the reconstruction interval is provided in Sec. II E. B. Proposed region-of-interest method The cine CT images were viewed on an image workstation Advantage Workstation 4.1; GEHealth Care, Waukesha, WI. To measure skin or tumor motions, a region of interest ROI was placed at the boundary of tissue/air/lung on the first CT image of a chosen axial slice cf. Fig. 2, and the same ROI same size and pixel coordinates was then propagated to the rest of the images for the same slice indexed by time. 15 The mean CT density in the ROI varied due to the motion of the anatomy. For each slice location, there was approximately one breath cycle worth of cine images for the motion analysis. Placement and size of ROIs for surface motion and tumor motion measurements are further discussed in Sec. II D. C. Validation We validate this ROI method by comparing the respiratory trace measured using this method with the RPM monitoring system. The RPM reflector block is usually not included in the 4DCT acquisition field of view for patients with lung cancers; therefore, to validate the ROI method, we used the cine CT dataset of a gastrointestinal case in which the RPM block was included in the scan. At the same axial slice, the external motions were measured with the ROI method and the RPM monitoring device. D. Selection of the axial slice and placement of ROIs 1. External chest wall motion To measure surface motion near the chest wall, we chose an 8-cm section of anatomy in axial length i.e., four couch positions near the location of the tumor for analysis. The eighth slice the most inferior slice from each of the four couch positions was selected for the chest wall motion analy-

3 3118 Chi et al.: External and internal motions 3118 FIG. 2. Placement of ROIs. To detect motion in the A-P and lateral directions, the ROIs were placed on the slice in which the center of the tumor can be observed during motion. To the right of the CT image is shown a tumor moving in the S-I direction, and the placement of the S-I ROI. sis such that chest wall motion was measured for four couch locations, at increments of 2 cm. Shown in Fig. 2, a region of interest of approximately 500 mm 2 was placed on the chest wall including some muscles under the skin on the first image, and the same ROI was propagated to all the images of the same slice location, indexed by time. The mean CT number in each ROI was recorded. 2. Tumor motion in S-I To measure tumor motion in the superior-inferior S-I direction, the axial slice which contained either the superior or inferior tip of the tumor, was chosen. The size of the ROI varied depending on the size of the tumor to ensure that the ROI was always slightly larger than the tumor width so that the mean CT density was not saturated throughout the cine data at this slice. Shown in Fig. 2, a tumor with a welldefined boundary moves in and out of the selected slice in the S-I ROI. Even though demonstrated in the same figure, the S-I ROI placement for tumor motion may not be on the same slice location as the ROI for chest wall motion had been. Figure 2 only demonstrates how the ROIs were placed with respect to the surrounding tissues. 3. Tumor motion in anterior-posterior/lateral To measure tumor motion in the anterior-posterior A-P or lateral directions, the axial slice at which the center of the tumor could be included was selected. The size of the ROI varied to make sure that the mean CT density would only vary due to the tumor motion, not other anatomy moving in and out of the ROI. The chosen slice location was the same for A-P and lateral motion measurements. E. Data analysis The relative phase differences were quantified between 1 the abdominal motion measured with the RPM and chest wall motion measured with the ROI method, 2 the abdominal motion measured with the RPM and tumor motion measured with the ROI method, and 3 the chest wall motion and tumor motion both measured with the ROI method. The RPM measurements and ROI measurements of mean CT numbers as functions of time were overlaid on the same plot. We compared the phase differences at 0% end inspiration and at 50% end expiration separately. The definitions of 0% and 50% for the motion of the chest wall and the tumor in the S-I, A-P, and lateral directions are listed in Table I. Since the tumor motion caused by the respiration is not predictable by location, size, and pulmonary function, 19 the definitions given in this study are according to the most likely tumor position at the end inspiration and end expiration for the purpose of phase shift analysis. The time differences for 0% and 50% were calculated between 1 four chest wall positions and the RPM, 2 tumor motion and RPM, and 3 tumor motion and chest wall motion. The time differences were next divided by the corresponding breathing cycle duration to obtain the relative phase shift in percentage. A positive relative phase shift in TABLE I. Definition of 0% and 50% based on the most likely direction of tumor motion. S-I, superior-inferior anatomic direction; A-P, anterior-posterior anatomic direction; LAT, lateral anatomic direction; ROI, region of interest. Percentage Abdominal motion observed by RPM Chest wall motion Tumor motion S-I A-P LAT 0 Highest amplitude end inspiration 50 Lowest amplitude end expiration Largest average CT number in ROI Smallest average CT number in ROI Tumor at most inferior position Tumor at most superior position Tumor at most anterior position Tumor at most posterior position Tumor farthest from mediastinum Tumor closest to mediastinum

4 3119 Chi et al.: External and internal motions 3119 TABLE II. Characteristics of the patients and 4DCT cine acquisition parameters. LUL, left upper lobe; RUL, right upper lobe: RL, right lobe; LL, left lobe; RLL, right lower lobe. Patient Patient s age yr Tumor location Distance of RPM block to tumor cm Image reconstruction interval s Avg. respiratory cycle duration s Uncertainty in ROI method 1 67 LUL RUL RL RL LL RLL RLL RL/RLL Liver patient for validation 69 N/A case 1 indicates that the chest wall motion was lagging behind the abdominal motion measured by the RPM, and a negative relative phase shift indicates that the chest wall motion was leading. The relative phase shift between chest wall and RPM measurements was reported as the average of the phase shifts of four couch positions. In case 2, a positive phase shift indicates that tumor motion was lagging behind the RPM. Similarly in case 3, a positive phase shift indicates tumor motion lagging behind the chest wall motion. Note that in case 3, the chest wall and tumor motions were measured from the cine data of the same respiratory cycle, and therefore at the same couch position. 1. Measurement uncertainty The measurement uncertainty in the ROI method of determining 0% and 50% phases was estimated for each patient by taking half of the image reconstruction interval divided by the averaged respiration cycle duration, and expressed in percentage as given in Table II. The uncertainty of respiratory trace measurement in the RPM monitoring system was negligible compared to the ROI method because the RPM system samples 30 data points per second, whereas approximately 15 cine CT image data points per respiratory cycle are available in the ROI method. The uncertainties in the phase shift calculations in cases 1 and 2 were the same as those given in Table II. The uncertainties in case 3 were estimated from the square root of summed uncertainties since both tumor motion and chest wall motion were measured with the ROI method. III. RESULTS AND DISCUSSION Characteristics of eight lung patients and one liver patient and the corresponding 4DCT cine acquisition parameters for this study are listed in Table II. A. Validation of the proposed ROI method Figure 3 shows that the waveform measured using the ROI method provided very similar respiratory waveforms to those given by the RPM method at the same slice location. The end inspiration 0% phases occurred at 0.27 and 4.27 s for the RPM measurement, and at 0.35 and 4.25 s for measurement with the ROI method. The differences are 0.08 and The end expiration 50% phases occurred at 3.30 and 6.04 s for the RPM measurement, and at 2.95 and 6.20 s for the ROI method with slightly larger differences of 0.35 and 0.16 s. The differences between RPM measurement and ROI method are caused by the lower sampling frequency in the ROI method 11 data points which could be improved by increasing the number of CT images per slice location, i.e., reducing the image reconstruction interval. B. RPM and chest wall motions Illustrated in Fig. 4 are respiratory traces for patient 2 measured with the RPM and the ROI methods. The amplitudes of the traces are only relative, not to be taken as the absolute displacements, and the tumor motion traces are intentionally offset in the y axis for the ease of viewing. The figure shows that chest wall motion at each couch position was lagging behind the RPM signal. It also shows that for the same respiratory cycle, the shape of the breathing trace observed on the surface of the chest wall can be different from that of the RPM measurement. This difference led to the relative phase shift at 0% being different from the phase shift at 50% as shown in Table III. For example, patient 4 showed an approximately 10% difference between the phase shifts, 0.8% at end inspiration, and 11.0% at end expiration. Other patients showed a smaller difference between the phase shifts at end inspiration and at end expiration. This difference can be caused by the combination of two factors: 1 the breathing waveform varies depending on the location of the measurement, and 2 each CT image is reconstructed from 0.5 s of projection data, therefore each image represents an averaged motion over 0.5 s. One solution for improving the resolution of the second factor is to use segment reconstruction. In a segment reconstruction, each image is reconstructed from 360 of projection data and therefore has better temporal resolution. The magnitude of the phase shift between chest wall motion and the RPM signal ranged from 1% to 31% at end

5 3120 Chi et al.: External and internal motions 3120 FIG. 3. Validation of the ROI method. The respiratory traces measured at the same slice locations by the RPM and the ROI methods. The reconstruction interval between images is 0.65 s. inspiration and from 11% to 24% at end expiration. The detailed results and their corresponding measurement uncertainties are presented in Table III. The standard deviations resulting from the four positions are larger than the uncertainty of the measurement, suggesting that small variations in the placement of the RPM block can lead to different RPM phase measurements. Abdominal motion preceded chest wall motion in seven of the eight patients, indicating that most of the patients in this study were abdominal breathers when lying in the supine position. Since different surface locations can lead to different respiratory trace measurements for the same respiratory cycle, as shown by the phase differences between abdominal motion and chest wall motion, and amongst four chest wall positions, it is important, when performing repeated studies, to place the RPM block at the same location as it was in the study to be compared. Although this action does not ensure the reproducibility of the breathing traces, it ensures that the errors due to variations in the RPM block placement are not introduced. C. Tumor motion and RPM/chest wall motion 1. Tumor motion and the RPM The relative phase shifts between tumor motion and the RPM signal motion are listed in Table IV. The result also showed that phase shift at the end inspiration could be different from phase shift at end expiration within the same respiratory cycle. For example, for S-I direction, we observed in patient 5 a 3%phase shift at end inspiration and a 16% phase shift at end expiration. This means that the external motion measured by the RPM reached its highest amplitude at approximately the same time as the tumor reached its most inferior position; however, at exhalation, external motion reached its lowest amplitude earlier by 0.5 s, a 16% shift than the tumor reached its most superior position. Similarly, for A-P direction, we observed in patient 6 a 15% phase shift at end inspiration and a 10% phase shift at end expiration. This indicates that A-P tumor motion reached its most anterior position after RPM reached its highest amplitude, and reached its most posterior position before RPM reached its lowest. Although we do not expect the phase shifts to be exactly the same, this large difference between the end inspiration 15% and the end expiration 10% could be caused by the uncertainty in the ROI method described previously. When comparing all patients, S-I tumor motion had significantly smaller relative phase shifts from the RPM than lateral and A-P motions had. Relative phase shifts in S-I motion ranged from 9% to 4% at end inspiration, and from

6 3121 Chi et al.: External and internal motions 3121 FIG. 4. Respiratory traces measured by the RPM and ROI methods. Xray on and off is shown as 1 and 0, respectively; no cine CT images are available for motion measurement during x ray off. Chest wall motions are studied on four slice locations, one slice location for each direction of tumor motion. The amplitudes of the traces are only relative, and not associated with true displacement. 13% to 16% at end expiration; phase shifts in lateral motion ranged from 15% to 54% at end inspiration and from 22% to 36% at end expiration; phase shifts in the A-P direction ranged from 1% to 48% at end inspiration, and 10% to 30% at end expiration. The observation of larger phase shifts in lateral and A-P directions could be caused by the combination of two reasons: 1 S-I motion dominates lateral and A-P directions in all eight patients, and 2 the nature of our proposed ROI method; since the ROI method only measured the motion in one slice location, the measurement is likely to favor the primary direction of the tumor motion, masking motions in the other two directions. In order to appropriately study the relation between external motion and internal tumor motion in A-P and lateral directions using the ROI TABLE III. Relative phase shift between chest wall motion and RPM measurement. The relative phase shifts are reported as the average of the four slice locations for each patient. Uncertainty is given as standard deviation SD of the four relative phase shifts. A positive phase shift means chest wall motion lags the RPM measured motion. End inspiration 0% End expiration 50% Patient no. Avg. relative phase shift SD Uncertainty Avg. relative phase shift SD Uncertainty

7 3122 Chi et al.: External and internal motions 3122 TABLE IV. Relative phase shift between tumor motion and the RPM signal. Uncertainties in the phase shifts were estimated from the measurement uncertainty in the ROI method. A positive phase shift means tumor motion lags the RPM signal. S-I, superior-inferior direction; LAT, lateral direction; A-P, anterior-posterior direction. End inspiration 0% End expiration 50% Patient no. S-I LAT A-P S-I LAT A-P Uncertainty method, tumors that move predominantly in A-P and lateral directions should be used. Another situation when masking motions could occur is if the ROI method is used on irregularly shaped tumors; in such cases, the CT mean density in the ROI can change even if no motion component in the analyzed direction is present. For this reason, only tumors with regular shapes were selected to be analyzed in this study. As a special case for phase shifts of approximately 50%, which could result from how the 0% and 50% are defined in Table I, the tumors may have moved opposite the most likely direction. 2. Tumor motion and chest wall The relative phase shifts between tumor motion and the chest wall motion are listed in Table V. A wide range of the phase shifts were observed in all three directions. No correlations were found between chest wall motion and the tumor motion in all three directions. TABLE V. Relative phase shift between tumor motion and the chest wall motion. Uncertainties in the phase shifts were estimated from the measurement uncertainty in the ROI method. Since both tumor motion and the chest wall motion were measured by the ROI method, the uncertainty is greater than it is in Table IV. S-I, superior-inferior direction; LAT, lateral direction; A-P, anterior-posterior direction. End inspiration 0% End expiration 50% Patient no. S-I LAT A-P S-I LAT A-P Uncertainty IV. CONCLUSIONS This study shows that the external motions measured by an RPM block placed near the abdominal region correlates well with tumor motion in the S-I direction. This result supports the placement of the RPM block near the abdominal region rather than the chest wall region because the majority of lung tumors move predominantly in the S-I direction. 20 A significant phase shift between the abdominal region and the chest wall region implies that for reproducibility it is necessary to keep the RPM block placed at the same axial location for simulation and treatment. To study the phase correlation between external surface motion and tumor A-P/lateral motion, we will continue with the ROI method by selecting patients whose tumors move predominantly in A-P or lateral directions and reduce the measurement uncertainty in the ROI method by reconstructing cine images at smaller time intervals. We will also investigate the potential of using 4DCT data sets to study the correlation between the external surface motion and tumor motion in its entirety. a Electronic mail: melinda.chi@di.mdacc.tmc.edu 1 P. J. Keall, V. R. Kini, S. S. Vedam, and R. Mohan, Potential radiotherapy improvements with respiratory gating, Australas. Phys. Eng. Sci. Med. 25, V. R. Kini, S. S. Vedam, P. J. Keall, S. Patil, C. Chen, and R. Mohan, Patient training in respiratory-gated radiotherapy, Med. Dosim 28, R. I. Berbeco, S. Nishioka, H. Shirato, G. T. Chen, and S. B. Jiang, Residual motion of lung tumours in gated radiotherapy with external respiratory surrogates, Phys. Med. Biol. 50, Y. Tsunashima, T. Sakae, Y. Shioyama, K. Kagei, T. Terunuma, A. Nohtomi, and Y. Akine, Correlation between the respiratory waveform measured using a respiratory sensor and 3D tumor motion in gated radiotherapy, Int. J. Radiat. Oncol., Biol., Phys. 60, A. Schweikard, H. Shiomi, and J. Adler, Respiration tracking in radiosurgery, Med. Phys. 31, H. Shirato, Y. Seppenwoolde, K. Kitamura, R. Onimura, and S. Shimizu, Intrafractional tumor motion: lung and liver, Semin. Radiat. Oncol. 14, S. S. Vedam, V. R. Kini, P. J. Keall, V. Ramakrishnan, H. Mostafavi, and R. Mohan, Quantifying the predictability of diaphragm motion during respiration with a noninvasive external marker, Med. Phys. 30, J. D. Hoisak, K. E. Sixel, R. Tirona, P. C. Cheung, and J. P. Pignol, Correlation of lung tumor motion with external surrogate indicators of respiration, Int. J. Radiat. Oncol., Biol., Phys. 60, E. C. Ford, G. S. Mageras, E. Yorke, and C. C. Ling, Respirationcorrelated spiral CT: a method of measuring respiratory-induced anatomic motion for radiation treatment planning, Med. Phys. 30, R. W. Underberg, F. J. Lagerwaard, J. P. Cuijpers, B. J. Slotman, J. R. van Sornsen de Koste, and S. Senan, Four-dimensional CT scans for treatment planning in stereotactic radiotherapy for stage I lung cancer, Int. J. Radiat. Oncol., Biol., Phys. 60, G. S. Mageras, A. Pevsner, E. D. Yorke, K. E. Rosenzweig, E. C. Ford, A. Hertanto, S. M. Larson, D. M. Lovelock, Y. E. Erdi, S. A. Nehmeh, J. L. Humm, and C. C. Ling, Measurement of lung tumor motion using respiration-correlated CT, Int. J. Radiat. Oncol., Biol., Phys. 60, E. Rietzel, G. T. Chen, N. C. Choi, and C. G. Willet, Four-dimensional image-based treatment planning: Target volume segmentation and dose calculation in the presence of respiratory motion, Int. J. Radiat. Oncol., Biol., Phys. 61, E. M. Leter, F. Cademartiri, P. C. Levendag, T. Flohr, H. Stam, and P. J. Nowak, Four-dimensional multislice computed tomography for determination of respiratory lung tumor motion in conformal radiotherapy, Int. J. Radiat. Oncol., Biol., Phys. 62,

8 3123 Chi et al.: External and internal motions D. A. Low, M. Nystrom, E. Kalinin, P. Parikh, J. F. Dempsey, J. D. Bradley, S. Mutic, S. H. Wahab, T. Islam, G. Christensen, D. G. Politte, and B. R. Whiting, A method for the reconstruction of four-dimensional synchronized CT scans acquired during free breathing, Med. Phys. 30, T. Pan, T. Y. Lee, E. Rietzel, and G. T. Chen, 4D-CT imaging of a volume influenced by respiratory motion on multi-slice CT, Med. Phys. 31, S. S. Vedam, P. J. Keall, V. R. Kini, H. Mostafavi, H. P. Shukla, and R. Mohan, Acquiring a four-dimensional computed tomography dataset using an external respiratory signal, Phys. Med. Biol. 48, E. Rietzel, T. Pan, and G. T. Chen, Four-dimensional computed tomography: image formation and clinical protocol, Med. Phys. 32, P. J. Keall, G. Starkschall, H. Shukla, K. M. Forster, V. Ortiz, C. W. Stevens, S. S. Vedam, R. George, T. Guerrero, and R. Mohan, Acquiring 4D thoracic CT scans using a multislice helical method, Phys. Med. Biol. 49, C. W. Stevens, R. F. Munden, K. M. Forster, J. F. Kelly, Z. Liao, G. Starkschall, S. Tucker, and R. Komaki, Respiratory-driven lung tumor motion is independent of tumor size, tumor location, and pulmonary function, Int. J. Radiat. Oncol., Biol., Phys. 51, H. H. Liu, B. Choi, J. Zhang, T. Tutt, P. C. Chi, C. Wang, D. Luo, K. Prado, R. Mohan, D. Lei, and P. Balter, in Assessing respiration-induced tumor motion and probability of internal target margin for image-guided radiotherapy of lung cancers, American Society for Therapeutic Radiology and Oncology ASTRO, Denver, CO, 2005.