Volume 33, Issue 6, June 2006
Index of content:
- Joint Imaging/Therapy Scientific Session: Valencia B
Therapy Localizatoin (Non‐Tomographic)
MO‐D‐ValB‐01: Characterization of Cardiac Motion in the Lung Using a Novel Electromagnetic System in An Animal Model33(2006); http://dx.doi.org/10.1118/1.2241413View Description Hide Description
Purpose: Previous studies have examined the accuracy of the use of three internal ACelectromagnetic transponders and wireless tracking system (Calypso® Medical) for tumor localization in prostate cancer. This study focuses on the use of the system to investigate and characterize cardiac induced lungtissue motion to better predict three‐dimensional lungtumor position in real‐time. Method and Materials: Under an institutional approved animal study, three 1.8 mm ACelectromagnetic transponders are bronchoscopically implanted in the periphery of the lungs of five hounds. The transponders are positioned in a triangle, each spaced 1–3 cm apart. The transponder positions are sequentially measured every 50 ms at five time points. During each measurement, the subject is stressed with several respiratory patterns. Signal processing of the data involves the design and application of a Butterworth highpass filter to obtain the component of transponder movement due to cardiac motion. Results: The data for the 1st three time points of the first animal are presented. FFT spectrum analysis indicated signal frequency components of 13.05 and 123.8 cycles/minute, due to respiration and cardiac motion respectively. Cardiac‐induced lungtissue motion was detected in vivo, ranging from 0.0007cm – 0.3592cm, by applying the highpass filter to the data. The motion was smaller on the implant day compared with the other two time points. Moreover, transponder position and distance from the heart had an effect on calculated motion. Finally, breathing patterns also affected the observed motion at a statistically significant 0.1% level. Conclusion:Cardiac contractions cause quantifiable motion in surrounding lungtissues that cannot be measured with existing onboard imaging capabilities. The motion varies depending on transponder position, distance from the heart, breathing pattern, and day of measurement. Though the motion maximum was 3.6mm, this motion could cause imaging artifacts when using respiratory correlates.
Research sponsored by Calypso® Medical Technologies.
33(2006); http://dx.doi.org/10.1118/1.2241416View Description Hide Description
Introduction. A limiting factor in registration of a three‐dimensional (3D) computed tomography(CT) dataset with two‐dimensional (2D) x‐ray images has been the time‐consuming generation of digitally reconstructedradiographs(DRRs). This can be overcome using a commercial graphics card for DRR generation enabling fast, robust, and accurate automatic image‐based 2D‐3D registration. Methods and Materials. For the iterative registration process hundreds of DRRs are created using hardware rendering in OpenGL. Each DRR from a 512×512×100 CT volume is rendered in less than 0.1 seconds using an nVidia 7800GT graphics card. The registration is based on a publicly available implementation of Mattes Mutual Information (ITK, U.S. National Library of Medicine, Bethesda, MD). To improve speed, the registration is performed on a sub‐image. A two‐step registration strategy is adopted for robustness with the first pass using a larger margin around the sub‐image and a down‐sampled resolution. A thoracic phantom (Model 602, CIRS, Norfolk, VA) was imaged and setup according to our clinical protocol and then shifted from 0–1.5 cm along each of the major axes. Anterior‐Posterior and Lateral kV x‐ray images were acquired using a commercial patient imaging system (OBI, Varian Medical System, Palo Alto, CA). Results: The mean registration times were 8 and 16 seconds without and with rotations respectively. We observed a systematic 1.1 mm offset in the longitudinal direction that we believe results from 2.5 mm CT slice spacing and OBI calibration. With this removed the mean three‐dimensional distance of the registered positions from the phantom positions was 0.4 mm with the largest disagreement being 0.75 mm. The systems ability to calculate rotations was only tested numerically.
Conclusions. The speed and accuracy of this system demonstrate that it could be a viable tool for reducing daily setup uncertainty by automating the analysis of setup images.
Research sponsored by Phillips Medical Systems.
MO‐D‐ValB‐03: Genetic Evolutionary Taboo Search: A Novel Approach for Optimal Marker Placement in Infrared Patient Positioning33(2006); http://dx.doi.org/10.1118/1.2241417View Description Hide Description
Purpose: To develop the methods of a novel approach for optimal marker placement in infrared patient positioning. Method and Materials: A nondeterministic optimization technique (Genetic Evolutionary Taboo Search, GETS), combining genetic algorithms and taboo search, was implemented. A population‐based evolution is generated, where adaptive memory features guide the evolutionary process to thoroughly explore the solution space. Preset taboo solutions are introduced to reject marker configurations resulting collinear from the point of view of infrared cameras. The GETS algorithm was tested on 10 prostate patients: treatment planningCT scans were segmented to provide 3‐D representation of PTV (prostate + seminal vesicles), OARs (bladder and rectum) and skin surface model. Segmented data were fed to the GETS algorithm to obtain optimized configurations of markers, minimizing the target registration error (TRE), to be compared to a random configuration. The changes in the optimal marker configuration when OARs are included within the target were also investigated. Results: The GETS algorithm yielded a significant improvement in TRE values: optimal configurations ensured a 26.5% mean TRE decrease. Common features in the optimal marker configurations were found for the 10 patients group, being optimized solutions symmetrically distributed, with markers mostly placed on lateral sides. Optimal marker configurations when OARs were included within the target resulted in a similar spatial distribution, if compared to the PTV‐only condition. The implemented memory‐based design resulted in improved gene expression over the evolution process, with respect to memoryless genetic algorithms. Conclusion: The GETS algorithm revealed high performance in solving the optimal marker placement problem, leading to improved marker configuration for stereophotogrammetric patient positioning in radiotherapy. Memory features ensured enhanced capabilities in exploring the solution space, if compared to conventional genetic optimization. The application of the new algorithm to a 10 patient group provided practical indications toward better marker placement for prostate cases.
MO‐D‐ValB‐04: Internal‐External Correlation Investigations of Respiratory Induced Motion of Lung Tumors33(2006); http://dx.doi.org/10.1118/1.2241418View Description Hide Description
Purpose: In respiratory‐gated treatments, the successful delivery of the planned dose distribution and sparing of the health tissue is highly dependent upon the assumption of a strong correlation between the external motion and the internal tumor motion. We will present a new internal/external correlation study based on a unique data set. Method and Materials: Radiopaque fiducial markers inside or near the target were implanted and visualized in real time by means of stereoscopic diagnostic x‐ray fluoroscopy. The fluoroscopic images were recorded continuously in synchronization with an external respiratory motion monitoring system. A data analysis methodology was developed in order to assess the correlation of the external breathing motion with the internal 3D position of the implanted fiducials. The methodology is based on a dynamic correlation technique and used to extract global correlation parameters as well as to reveal their instantaneous behavior. Results: We have found that in some cases, the poor internal/external correlation is caused by a time mismatch between the motion of the internal fiducial markers and the external breathing motion. For some cases, there is a sizeable time delay between the internal tumor motion and the external motion of up to 0.8 seconds, revealing that internal‐external motion coupling is dependent on the tumor position. We have also found that the time delay itself is time‐dependent. Conclusion: The proposed technique reveals one of the causes for poor internal‐external correlation and it could be used to improve the current gated treatment methodology by combining the amplitude gating technique with the measured time‐delay. In the course of these investigations, we also found that our technique can reveal difficulties in extracting the underlying time delay (due to its own time dependence) and that one has to be careful of how the time delay is implemented for gating.
33(2006); http://dx.doi.org/10.1118/1.2241419View Description Hide Description
Purpose/Objective: To establish a commissioning and acceptance test protocol for an ACelectromagnetic tracking system for use in target localization and tracking during radiation therapy. (Calypso® 4D Localization System, Calypso Medical, Seattle, WA). Materials/Methods: Following installation of infrared cameras and electromagnetic system, compatibility tests were made to ensure the system did not interfere with radiation output, MLCs or IMRT field fluence. The collision space between the linac and the Calypso System was evaluated. System verification included calibration using software and calibration tools, one of which has embedded RF transponders and optical reflectors. End‐to‐end testing to assess localization accuracy included CT scan of a radiographic phantom with RF transponders, creation of four Calypso Plans, data entry to the Calypso System and execution of treatment sessions. Tracking accuracy was measured using a precision translation table for orthogonal motions ± 5 mm. Results: Calypso System did not affect normal clinical operations of radiation output, MLC, or IMRT field segmentation. Overall accuracy on ten systems at five institutions and 40 treatment plans was 0.068 ± 0.027 cm, incorporating contributing errors from the CT scans, identification of transponders in the phantom, and effects of radiation dose delivery. With stable environment, no systematic drift was noticed over 30 minute. Translations of precise increments from −0.50 cm to +0.05cm were accurately tracked with error of 0.00cm, −0.02cm and 0.01cm (lat, long, vert). The system maintained accuracy with monthly optical calibrations. Changes in system readout of ±0.05 cm (readout quantization) were noticed at certain gantry angles. Conclusion: Evaluations demonstrate the ACelectromagnetic system with wireless transponders can be integrated into the radiotherapy environment with existing instrumentation and operates within the designed accuracy specification. Conflict of Interest: Work supported by Calypso Medical Technologies.
MO‐D‐ValB‐06: Concurrent Tracking and Fluoroscopic Imaging of Implantable Wireless Electromagnetic Transponders33(2006); http://dx.doi.org/10.1118/1.2241420View Description Hide Description
Purpose: Multiple technologies are being utilized to improve real‐time tumor tracking. To date, there have not been methods to prospectively compare different technologies with realistic tumor trajectories. We evaluated the capabilities of the Calypso® Medical 4D Localization System (Calypso Medical, Seattle, WA) and Varian Trilogy System (Varian Medical Systems, Palo Alto, CA) fluoroscopy in tracking dynamic objects. Method and Materials: Initially, a quality assurance fixture containing three implantable transponders was moved by an in‐house developed 4D phantom through an ellipse and a non‐uniform human lungtumor path modeled with CTimaging and spirometry. Subsequently, three transponders that had been implanted in a canine lung were tracked. In both experiments, the transponders were fluoroscopically imaged on a Trilogy system while simultaneously being tracked by the Calypso® 4D localization system. The fluoroscopic images were recorded and later analyzed using a custom‐written (MATLAB) image processing program to determine the transponder projection positions with respect to time. The trajectories derived from the fluoroscopic images were synchronized with and compared to the Calypso System position data. Results: The root mean square (RMS) position differences were less than 0.03 mm for all tested measurement system combinations. While both were small, the Calypso System RMS error was slightly lower than that of the fluoroscopy when compared against the 4D phantom positions. Of the three trajectories, the RMS error between imaging modalities was largest for the patient trajectory and smallest for the ellipses. Conclusion: This work indicates that both tracking methods provide excellent positioning accuracy. Although the accuracy discrepancy between the two systems is negligible, the Calypso® System also offers the ability to localize in three dimensions and has the advantage of being able to track a target continuously without the use of ionizing radiation.Conflict of Interest: Supported in part by Calypso Medical Technologies, Inc.
MO‐D‐ValB‐07: Comparison of Inline and Orthogonal Imaging and Treatment Beam Geometries for Monitoring the Motion of Implanted Markers33(2006); http://dx.doi.org/10.1118/1.2241421View Description Hide Description
Purpose: The purpose of this study is to investigate the accuracy of different 2D methods monitoring implanted markers, compared to 3D method for real‐time tumor tracking radiotherapy. The different imaging‐treatment beam geometries were the imaging beam parallel to treatment beam (inline) and orthogonal to treatment beam.
Method and Materials: The 3D motion datasets of ten patients from Cyberknife treatments were used. For given beam angles, the positions of implanted markers were calculated and its geometric uncertainty was quantified for two 2D monitoring methods. Since neither can monitor the motion of markers in the imaging beam axis, the geometric errors were determined in that direction with respect to the treatment beam. Assuming that 3D pre‐treatment online positioning was performed and thus errors are predominantly random, treatment margins due to the limitations of 2D methods were quantified. For the orthogonal monitoring, margin, M=1.65σ, was used with the assumption of zero systematic errors; while for the inline, the required margin can be calculated by integrating the probability density function of the geometric uncertainty with the dose falloff along the beam direction. Results: In terms of couch motion, the positioning uncertainty is lowest for coplanar treatments, consistent with predominantly superior‐inferior motion. Regarding gantry angles, it is lowest for lateral beams, consistent with the smallest left‐right motion. The average positioning uncertainty along the imaging beam axis is 0.05–0.16cm (1 SD) with maximum values for individual patients ranging 0.09–0.33cm, which result in 0.08–0.26cm margins for orthogonal monitoring.Conclusion: The impact of the geometric relationship between the imaging and treatment beam has been studied by quantifying the error from out of plane motion for inline and orthogonal imaging‐treatment geometries. The errors for inline geometry result in negligible additional margin required. In the absence of other errors, the orthogonal monitoring contributes up to 0.26cm to the total margin.
33(2006); http://dx.doi.org/10.1118/1.2241422View Description Hide Description
Purpose: To develop techniques for direct lungtumor tracking in fluoroscopic images without implanted markers. Method and Materials: During the patient setup session, a pair of 15 second orthogonal fluoroscopic images are taken and processed off‐line to generate reference templates. Each breathing cycle is divided into 12 phase bins. Setup image frames falling in a specific bin are motion‐enhanced and averaged, and an ROI that contains the tumor is selected to be the reference template for that phase bin. Each reference template corresponds to a tumor position in the image. During the treatment, as soon as a fluoroscopic image is acquired, the cross‐correlation score between each reference template and this image is maximized by allowing small shifts of the template in both X and Y directions. Then the tumor position is derived by averaging the tumor centroid coordinates in those templates of high scores (above 85% of the maximum score). For comparison, tumor position in each image frame was also marked by a clinician. Results: We tested our algorithm on six sequences of fluoroscopic images from six lungcancer patients. The automatically detected tumor centroid coordinates agree well with the manually marked results, with an average error of 1 mm. Conclusion: This study demonstrates the feasibility of tracking lungtumor mass in fluoroscopic images without implanted fiducial markers. Future research will concentrate on further improvement of the accuracy and robustness, and reducing the computational cost.
The project is partially supported by NCI grant (1 R21 CA110177 A 01A1) and NSF Grant No. IIS‐0347532.
33(2006); http://dx.doi.org/10.1118/1.2241423View Description Hide Description
Purpose: The intention of this study is to determine whether previously observed large external surrogate residual motion at end‐or‐inhale (EOI) translates into large tumor residual motion, and if improving the reproducibility at this phase can lessen the internal residual motion. Method and Materials: We simulate gated treatment at the EOI phase, using a set of recently measured internal/external correlated patient data. The 3D locations of internal fiducial markers placed near the target are tracked in real‐time with stereoscopic x‐ray fluoroscopy. An external surrogate respiratory gating system is synchronized with the fluoroscopic unit so that the log files contain the three‐dimensional marker position and the abdominal surface position at every time point. The internal and external measurements are taken even when the MV beam is gated off, throughout each treatment, so large amounts of internal/external‐correlated data were collected. Results: We found that under free‐breathing conditions the residual motion of the tumors is much larger for EOI phase than for end‐of‐exhale (EOE) phase. The mean value of residual motion at EOI was found to be 2.2 mm and 2.7 mm for amplitude and phase‐based gating, respectively; and, at EOE, 1.0 mm and 1.2 mm for the same quantities. However, the residual motion in the EOI gating window is correlated well with the reproducibility of the external surface position in the EOI phase. Using the results of a published breath‐coaching study, we deduce that the tumor residual motion at EOI would approach that at EOE under breath‐coaching conditions. Conclusion: We conclude that the same reproducibility of tumor location can be achieved at EOI as at EOE if breath coaching is implemented. Based on these results, we believe that inhale gating is preferable to exhale gating as long as proper margins are employed and breath coaching is performed.
Margin Assessment and Modeling of Inter‐Fraction Motion
TU‐C‐ValB‐01: Evaluation of Clinical Margins Via Simulation of Patient Setup Errors in 27 Prostate IMRT Plans33(2006); http://dx.doi.org/10.1118/1.2241513View Description Hide Description
Purpose: To evaluate: (i) the size of random and systematic setup errors that can be absorbed by 5mm CTV‐to‐PTV margins in prostate IMRTtreatment plans; (ii) whether findings are consistent with published margin recipes; (iii) if shifting contours with respect to a static dose distribution accurately predicts dose coverage due to setup errors.
Method and Materials: 27 IMRTtreatment plans with 5mm CTV‐to‐PTV margins were utilized. Random setup errors with standard deviations (SDs) of 1.5, 3 and 5mm were simulated by fluence convolution. Systematic errors with the same SDs were simulated using two methods: (a) shifting the isocenter and recomputing dose (isocenter shift), and (b) shifting patient contours with respect to the static dose distribution (contour shift). Maximum tolerated errors were evaluated such that 90% of plans had target coverage greater than a specified minimum. Results: For coverage criteria consistent with margin formulas, plans generated with a 5mm margin were able to absorb SDs >3mm. Most structures, including the prostate CTV, showed close agreement between isocenter and contour shift methods. Exceptions were the nodal CTV and small bowel. For 3mm SDs, contour vs isocenter shift estimates for the percent of plans with acceptable dose differed by >2% for the nodal CTV, and >7% for the small bowel. Contour shift small bowel D30 values differed from isocenter shift values by >120% for some simulated shifts. Conclusion: Published recipes require margins of 8–10mm for 3mm SDs. For the IMRT cases presented here, a 5mm margin would suffice. Approximating structure doses by shifting contours with respect to a static dose distribution was acceptable for most structures, but resulted in significant errors for the nodal CTV and small bowel doses for some shifts due to proximity to high dose gradients. (Work supported by NIH R01CA98524).
TU‐C‐ValB‐02: Patient Specific Differences in Setup Error Variability and Its Effect On Treatment Margins in Fractioned Radiotherapy33(2006); http://dx.doi.org/10.1118/1.2241514View Description Hide Description
Purpose: It is often assumed that geometric error distributions in radiotherapy differ from patient to patient. It is however, problematic to substantiate this assumption because, generally, limited measurements are available per patient, giving a high uncertainty in the estimate of the standard deviation (SD). Our aim is to develop a simple method to estimate the true distribution variability based on statistical data analysis of a large patient population and to investigate the effect of detected variability on population based PTV‐margins. Method and Materials: Setup error data of 470 prostate cancer patients (11 portalimaging measurements per patient, used for off‐line corrections), were analyzed for random errors. The SD of the setup error was computed for each patient. The RMS‐values of these numbers estimate the random uncertainty in the patient population. Next, the SD of the SD per patient is computed, containing the real distribution variability diluted by “measurement error in the SD” due to the limited number of samples. To estimate the true distribution variability, a correction is applied for this “measurement error”. Finally, that margin was calculated that encloses the CTV with the 95% isodose for 90% of the population. Results: The true inter‐patient variability is 26% of the SD, found after correcting for a “measurement error” of 18% (11 samples). Inter‐patient distribution variability leads to larger PTV‐margins, partly because the range of dose blurring becomes patient dependent. Assuming normality and the same SD variability in random and systematic errors, the margin for systematic errors increases from 2.5SD to 2.8SD, maintaining the same margin for random errors. Conclusion: Inter‐patient distribution variability exists but only slightly exceeds its measurement error and it is therefore difficult to detect for individual patients. By grouping many patients, it can be detected. A variable distribution requires slightly larger margins than a homogenous one.
33(2006); http://dx.doi.org/10.1118/1.2241515View Description Hide Description
Purpose: To quantify the dosimetric impact of intrafractional motion on reduced‐margin IMRT treatments of prostate cancer.Methods and Materials:CTimages were acquired immediately before and after a daily treatment for 46 prostate cancer patients. These CT sets were registered to the bony anatomy of the patient using an in‐house 3D image registration software. To test the hypothesis that a 3‐mm isotropic target margin would adequately cover the target over the duration of the treatment, an 8‐field IMRT plan was designed on the pre‐treatment CT and subsequently copied and re‐calculated on the post‐treatment CT. For convenience of comparison, dose plans were designed to receive a full course of treatment (75.6Gy). Dosimetric impact was assessed with comparisons of prostate, seminal vesicle (SV), rectum, and bladder volumes receiving several dose levels as well as the minimum and maximum doses to 0.1cc of the prostate and SV. Anatomic variations were also quantified. Results: Over the duration of one treatment fraction (21.4+/−5.5 minutes), there were systematic reductions in the volumes of the prostate and SV receiving the prescription dose (1.8 and 7.2 % respectively, P<0.001) as well as the minimum dose to 0.1cc of their volumes (2.1 and 6.4Gy, P<0.001). Of the 46 patients, 4 patients' prostates (91%) and 8 patients' SVs (83%) did not maintain dose coverage above 70Gy. Rectal dose increased and dose to the percentage‐volume of the bladder decreased at all dose levels. Rectal volume filling was correlated with a decrease in percentage‐volume of the SV receiving 75.6, 70, and 60Gy (P<0.001, P<0.001, P=0.02). Conclusion: With a 3‐mm intrafractional margin, a considerable percent of patients will not receive full dose coverage. The rectal volume increase during a treatment fraction has significant dosimetric impact on SV dose coverage and rectal sparing. Proactive immobilization of the rectum during treatment may be warranted.
TU‐C‐ValB‐04: Margin‐Less Prostate IMRT Plans, Directly Optimized for TCP and NTCP Including Geometric Uncertainties33(2006); http://dx.doi.org/10.1118/1.2241516View Description Hide Description
Purpose: To account for geometric uncertainties without the use of margins during IMRT planning such that optimal values are obtained for the population averaged TCP and NTCP functions. Methods and materials: A new method of computing cost functions was implemented within the IMRT planning tool Hyperion. Population‐averaged values of biologic score functions (TCPpop and NTCPpop) are optimized, simulating random errors by blurring the dose, and systematic errors by displacing target and OARs relative to the dose distribution.
For 19 prostate (and seminal vesicle) patients, treatment plans for a five beam setup were created, optimising TCPpop while constraining rectum NTCPpop and the maximum dose to the target. Gaussian distributions were used for the systematic and random errors (translations only, no attempt was made to model rotations or deformations). Since geometric uncertainties were accounted for within the cost functions, no CTV to PTV margin was used. For comparison, conventional plans were created using a CTV‐to‐PTV margin (M=2.5Σ+0.7σ) and a Simultaneous Integrated Boost (SIB) technique (68Gy to the above PTV, 78Gy to PTVboost with 5mm margin, 0mm towards rectum). The resulting plans were evaluated using an independent tool that simulates the effects of geometric uncertainties. Results: Compared to conventional plans, our new technique reduced the planned dose to the rectum, while increasing the volume receiving 78Gy. We ensured that TCPpop of the new technique was not smaller than for conventional techniques. The average rectum NTCPpop values were 14% (margin recipe), 8% (SIB), and 4% (new technique), for average TCPpop values of 69%, 70%, and 71%. Conclusions: The computation of TCP and NTCP including knowledge of geometric uncertainties within the inverse IMRT optimization loop is feasible (less than 1 hour optimization time), and results in robust prostate treatment plans with an improved balance between local control and rectum toxicity.
33(2006); http://dx.doi.org/10.1118/1.2241517View Description Hide Description
Purpose: The purpose of the present work is to assess the changes in size and respiration‐induced motion of lungtumors resulting from radiation treatment.Methods and Materials: Six to ten four‐dimensional computed tomography (4‐DCT) image datasets were acquired for each of 5 stage‐III non‐small‐cell lungcancer patients who received chemoradiotherapy treatment over six weeks. Serial 4‐D datasets were obtained each week. Gross tumor volumes (GTV) were outlined on each data set. Software tools in the radiation treatment planning system were used to calculate the volumes and centroids of the GTVs on the 0% (end‐inspiration) and 50% (end‐expiration) phase for each dataset. Interfractional changes in GTV location was assessed by registering corresponding phases of the datasets based on vertebral body landmarks and determining variations in the position of the GTV centroids relative to the landmarks. Forty‐six scans including six primary tumors (involved nodal stations were not included) were analyzed. Results: The initial mean tumor volume was 53 cm3 (range: 1 to 137cm3). The interfractional changes in GTV position were predominantly in the superior‐inferior direction with a mean magnitude of 3.4mm (range: 0.1 to 9.3mm). Overall tumor regression ranged from 20–71% (0% phase) and 15–70% (50% phase). As tumors shrunk, the magnitude of intrafractional GTV motion increased in the anterior‐posterior and superior‐inferior directions while remaining constant in the right‐left direction. Reproducibility of the GTV‐centroid position at the 50% phase, based on same‐day repeat CT scans, was within 2 mm in each direction. Conclusions: Because of changes in tumor size and intrafractional tumor motion, care must be taken when reducing treatment portals based on explicit determination of the internal target volume (ITV). Repeat 4‐DCT scans might be warranted during treatment.
TU‐C‐ValB‐06: Intra‐Fraction Motion of Immobilized Intra‐Cranial and Extra‐Cranial Patients Assessed by the CyberKnife Image‐Guidance System33(2006); http://dx.doi.org/10.1118/1.2241518View Description Hide Description
Purpose: To quantify intra‐fraction motion of immobilized intra‐cranial and extra‐cranial patients. The data can be used to optimise the intra‐fraction imaging frequency and consequent patient set‐up correction with the CyberKnife image‐guidance system and to establish the required margins in the absence of such a system. Method and Materials: We analysed the intra‐fraction motion of 21 intra‐cranial patients, who were immobilized with a thermoplastic mask and 9 supine and 8 prone treated extra‐cranial patients, who were immobilized with a vacuum bag. The motion was recorded by the CyberKnife image‐guidance system. We analysed the intra‐fraction motion by calculating the mean displacement with the standard deviation (SD) as a function of the time between kV X‐ray localizations. For the three groups separately, we calculated the systematic (overall mean and SD) and the random displacement as a function of the imaging frequency. Results: For all patients, the overall mean displacement was below 0.5 mm (3D vector) over a period of 15 min and hardly increased. The SD of the systematic displacements increased linearly over time for all 3 patient groups. For intra‐cranial, supine and prone treated patients, this SD increased to 0.5, 1.2, and 1.6 mm, respectively, in a period of 15 min. The random displacements for the prone treated patients were significantly higher than for the other groups, namely 1.3 mm (1 SD). This was most pronounced in the AP direction, suggesting that the larger intra‐fraction motion was caused by respiratory motion. Conclusions: Repeated intra‐fraction imaging and consequent patient set‐up correction with an interval of less than 5 min adequately compensates for patient motion during treatment. In the absence of this procedure, intra‐fraction motion has to be accounted for in the PTV margin.
33(2006); http://dx.doi.org/10.1118/1.2241519View Description Hide Description
Introduction: Our objective was to characterize retrospectively acquired 4DCT data for prospective gated delivery, and the effects of gate length on beam energy stability, output constancy, and positional accuracy / inter‐device constancy. Materials and Methods: A barometric sensor gated the Siemens Oncor linac and Siemens Sensation CT scanner. Respiratory motion of 20 mm at 15 bpm over a stationary jig was used to assess radio‐opaque marker positions. Retrospective 4DCT reconstructions were obtained at 6 phases of inspiration and expiration, ranging from 0% to 100% by 20% intervals. The center of the pin was identified using 50% threshold values on the CT dataset. On the linac, gate windows of 1500, 850, 500, 350, 300, and 250 ms for the 12 phases were studied. Ion chambers were used to measure the beam energy and output stability at 10 cm and 20 cm in solid water simultaneously. Marker position during gated delivery was determined via film. Nine profiles, centered around the marker, were extracted for both the static and moving axes. The averages were smoothed, and the peak position and full‐width‐at‐half‐maximum (FWHM) were determined. The difference in FWHM along the static and moving axis is the intra‐gate motion. Results:Dosimetry for gates >= 500 ms was excellent. Although the average energy was constant, gate length reduction from 500ms to 250 ms resulted in an energy standard deviation increase of 1–14%, an output constancy increase from 1.6% to 4.6%, and a 50% dose rate decrease. Mean discrepancies between marker position measured on CT and linac were 3 mm, with 8 mm maximum. Conclusions: Dosimetric characteristics of the linac are reasonable for gating windows ⩾ 500 ms. Target position measured on retrospective 4DCT can introduce significant uncertainty for several phases of the respiratory cycle.
33(2006); http://dx.doi.org/10.1118/1.2241520View Description Hide Description
Purpose: To evaluate the impact of internal organ variations on IMRT treatments of head‐and‐neck cancers using different daily alignment techniques. Method and Materials: Eleven head‐and‐neck cancer patients were imaged twice weekly during their course of treatment (141 CT scans total) using an integrated CT‐linear accelerator (EXaCT, Varian Oncology Systems). The clinical IMRT plans were copied onto the daily CTimages. The plans were aligned with (1) the daily marked isocenter using three radio‐opaque markers (BBs) and (2) bone alignment, using in‐house software to align the cervical vertebrae. Daily dose distributions were mapped from the daily CTimages onto the planning CTimage with an in‐house deformable image registration algorithm. Cumulative dose‐volume histograms from the planning CTimage were analyzed.Results: The differences in the clinical target volumes (CTV) gEUD between the planned and delivered doses (with BB or bone alignment) were typically ⩽1Gy; therefore the differences in target coverage were most likely clinically insignificant. However, the alignment method did have a statistically significant impact on the percentage‐volume of the CTV at the prescription dose. Neither BB alignment nor bone alignment maintained the planned coverage (average=98.2%), which was reduced to 95.6% with bone alignment (p=0.000) and 94.3% with BB alignment (p=0.000).
BB alignment significantly increased the average percentage‐volume receiving ⩾25Gy above the original plan for the ipsilateral (59.6% vs. 51.4%, p=0.003) and contralateral (42.0% vs. 36.4%, p=0.016) parotid glands. The parotid gland gEUD increased by more than 5Gy in 35% of BB alignments and 15% of bone alignments. However, there was no statistically significant difference between BB and bone alignments in parotid dose received. Conclusions: The differences in CTV coverage between bone and BB alignment were statistically significant but small. Bone alignment more closely reproduced the planned parotid dose than BB alignment, although both gave higher dose than the original plan.
TU‐C‐ValB‐09: Setup Error Analysis of HN‐IMRT Patients Using Electronic Portal Images and Cone‐Beam CTs33(2006); http://dx.doi.org/10.1118/1.2241521View Description Hide Description
Purpose: It is important to monitor and correct patient setup during treatment course for head and neck IMRT because highly conformal dose distribution is sensitive to setup uncertainties. Setup for HN region is unstable because patient is usually uncomfortable under the mask and the flexible bony structures in the neck region. The purpose of this study is to analyze the setup errors during entire treatment course. These findings will help make appropriate corrective decisions. Method and Materials: Patients enrolled in our IMRT protocol are immobilized with a large thermoplastic mask attached to the MedTec IPPS. 2D analysis is accomplished by comparing electronic portal images to DRRs using in‐house software. Systematic setup error exceeding 3mm is corrected. 3D analysis is performed by registering cone‐beam CT to planning CT. Data from 21 patients with total 185 sessions were used. Correlation between 2D and 3D were analyzed. Time trend was analyzed for patients with daily CBCTs (4 patients with 131 scans total). Results: Good correlations were observed between 2D and 3D analyses with mean difference less than 1mm. Both methods showed that the mean of setup errors is under 1mm in all directions. The systematic and random errors were about 2mm. Margin of 5mm used in the planning seemed to be adequate based on empirical recipes. Time trend analysis shows that changes occurring during treatment course are significant for 3 (out of 4) patients. Conclusion: 2D and 3D analyses agree with each other, but 3D should be used whenever possible because it has the advantage of better image quality, lower imaging dose, and better software to interpret information. The difference is caused mainly from image quality and non‐rigid bony motion. It may be necessary to redo the mask in the middle of treatment course to reduce overall setup error.
33(2006); http://dx.doi.org/10.1118/1.2241522View Description Hide Description
Purpose/Objective: To investigate the change in rectal dose during the treatment course for prostate IMRT with image‐guidance.Materials & Methods: Ten prostate cancer patients treated with IMRT were included in this study. Each patient was administered an enema prior to CT‐ and MRI simulations. MR and CT were fused for target delineation. IMRTtreatment planning was performed on the CTimage. Prostate motion during the treatment course was corrected using a CT‐on‐Rails system. Rectal contours were generated on both simulation CTimages and subsequent treatmentCTimages.IMRT plans were generated based on our clinical acceptance criterion. The subsequent treatmentCTimages for each patient from the CT‐on‐Rails system were used to recompute the patient dose distributions with the same leaf sequences used for treatment. The isocenter was shifted relative to the simulation CT, as required by the protocol, to ensure appropriate target coverage. The rectal doses based on the subsequent treatmentCT were compared with the original doses planned on the simulation CT scans using our clinical acceptance criteria. Results: Based on ten patients with 84 treatmentCT sets, 14% of the subsequent treatmentdose distributions did not meet our criterion of V40 < 35% (V40=36%∼50%), and 7% did not meet our criterion of V65<17% (V65=18%∼36%). The inter‐fractional rectal volume variation is significant for some patients. The minimum changes in rectum volume are between 31 and 39.8cc while the maximum changes are between 50.2 and 161.7 cc. In general, IMRT planning with an empty rectum results in better subsequent treatmentdose distributions to the rectum. Conclusions: Due to the large inter‐fractional variation of the rectal volume it is more favorable to plan prostate IMRT based on an empty rectum.
Molecular Imaging & Image Registration / Fusion
WE‐D‐ValB‐01: Effects of PET Reconstruction Parameters On the Delineation of Heterogeneous Target Volumes33(2006); http://dx.doi.org/10.1118/1.2241754View Description Hide Description
Purpose As emerging radiotherapy techniques incorporate biological targeting of sub‐tumor volumes, steps must be taken to ensure the validity of the assumed substructures. This study measures the effects of PET image reconstruction on heterogeneous target definitions both in vivo and in a phantom. Method and Materials: A known heterogeneous phantom composed of Y‐86 and Ge‐68 spheres with an F‐18 background altering signal‐to‐background ratios tested the accuracy of reconstructions using ordered subset expectation maximization (OSEM) with varying numbers of iterations and filtered backprojection (FBP) with Hanning, Shepp‐Logan, and ramp filters. In vivo measurements used heterogeneously proliferating tumorimages obtained from a canine tumorimaged using [F‐18]FLT at three stages of treatment using the same reconstruction methods. Difference images and standard deviations were used to assess the reconstruction differences. A three‐dimensional form of the Moran I(d) spatial statistic was used to assess global heterogeneity at various correlation distances.
Results: Absolute difference images from FBP and 2 iteration OSEM reconstructions showed internal tumor voxel clusters deviating by more than 10% of the maximum SUV of the reference image (OSEM20) and relative voxel values varying by as much as 40% in tumor periphery. Image differences in OSEM reconstructions significantly decreased after 10 iterations, accompanied by decreases in the standard deviation of differences and slight increases in heterogeneity as global I(d) values decreased. FBP reconstructions both underestimated (Hanning, Shepp‐Logan) and overestimated (ramp) global heterogeneity I(d) relative to reference values, but large standard deviations of absolute difference indicated images compared poorly to the reference. Conclusion:Tumor heterogeneity obtained through PET may vary by at least 10% internally with larger variability at the periphery, greatly affecting both tumor volume delineation and internal heterogeneity. Prescriptions for dose painting based on proliferation measures can vary widely with the reconstruction algorithm.