B Yi

SU-E-J-265: Practical Issues and Solutions in Reconstructing and Using 4DCT for Radiotherapy Planning of Lung Cancer

PURPOSE To report practical issues and solutions in reconstructing and using 4DCT to account for respiratory motion in radiotherapy planning. METHODS Quiet breathing 4DCT was used to account for respiratory motion for patients with lung or upper abdomen tumor. A planning CT and a 4DCT were acquired consecutively with a Philips Brilliance CT scanner and Varian RPM System. The projections were reconstructed into 10 phases. In Pinnacle RTP system, we contour a GTV in each phase and unite all 10 GTVs as ITV. The ITV is then mapped to the planning CT. We describe practical issues, their causes, our solutions and reasoning during this process. RESULTS In 6 months, 9 issues were reported for 8 patients with lung cancer. For two patients, part of the GTV (∼50% and 10%) in planning CT fell outside the ITV in 4DCT. There was a 7 mm variation in first patient back position because less restricted immobilization had to be used. The second discrepancy was due to moderate variation in breathing amplitude. We extended the ITV to include the GTV since both variations may likely happen during treatment. A LUL tumor showed no motion due to a 10-s long no-breathing period. An RLL tumor appeared double due to an abnormally deeper breath at the tumor region. We repeated 4DCT reiterating the importance of quiet, regular breathing. One patient breathed too light to generate RPM signal. Two issues (no motion in lung, incomplete images in 90% phase) were due to incorrect tag positions. Two unexplainable errors disappeared when repeating reconstruction. In summary, 5 issues were patient-related and 4 were technique issues. CONCLUSION Improving breathing regularity avoided large artifacts in 4DCT. One needs to closely monitor patient breathing. For uncontrollable variations, larger PTVs are necessary which requires appropriate communication between physics and the treating physician.

SU‐E‐T‐38: Are the Calculation Methods for Determining Tissue‐Maximum Ratios from Percent Depth Dose Valid for Flattening Filter‐Free Photon Beams?

Purpose: The calculation methods used to determine tissue‐maximum ratio (TMR) from percent depth dose (PDD) were derived by considering differences between TMR and PDD such as the geometry and field size. Phantom scatter factors or peak scatter factors are used to correct dosimetric variation due to field size differences. This relation matches well when the photon beam is flat. However, validity of this equation for flattening filter‐free (FFF) photon beams has never been evaluated. This study aims to evaluate the validity of the equation for FFF beams. Methods:Measured TMRs of 6FFF and 10FFF beams (TrueBeam, Varian) were compared to calculated TMRs. The TMRs were calculated from PDDs using scatter factors both from BJR Supplement 25 and from our own measurements. The same methods were repeated for standard flat 6MV and 10MV beams to confirm the soundness of the measurements and calculations. Results: for the flat beams, the calculated and measured TMRs agree within 1.2% over all field sizes and depths. For the FFF beams, differences are minimal up toa depth of 10 cm for all field sizes. For field sizes less than 20 cm, no noticeable differences are found down to 30 cm depth. The calculated and measured TMRs start to deviate around 15 cm depth for field sizes larger than 20 cm. Differences increase with depth and field size, resulting indeviations up to 3.2%. Conclusions: For field sizes smaller than 20 cm and depths shallower than 15 cm, traditional calculation models for TMR are acceptable for FFF beams; however, beyond those limits, the models start to deviate from measurement. We believe this is due to decreased lateralscatter of the FFF beam. For clinical FFF beams, TMR should either be measured directly, or a new calculation model should be developed.

MO-F-BRC-05: Minimization of the Total Inter-Segment Time for the Real-Time Tumor Tracking with Step & Shoot IMRT

Purpose: We developed an algorithm to minimize total inter‐segment time (TIST) for the MLC‐based real‐time tumor tracking with step‐and‐shoot IMRT. This algorithm optimizes a starting phase of tumor motion for each segment to minimize TIST. Methods: The optimizing algorithm consists of four steps: (1) implementation of feathering motion for the closed leaves that will be opened at the next segment, (2) calculation of inter‐segment time for all segments, (3) reordering segments to minimize TIST, and (4) optimization of the starting phase of tumor motion for each segment to minimize TIST. Thirty step‐and‐shoot IMRT fields from five patients with lung and abdominal cancer were used to test the algorithm. Tumor motion was varied with a period (2.0 to 4.0 s) and a peak‐to‐peak distance (0.5 to 4.0 cm). TIST and duty cycle for each field were compared to those from the strategy of starting each segment at end‐of‐exhale. Results: The TIST was reduced by 54.0% on average (from 30.2 ± 16.9 to 13.9 ± 10.6 s) and, the effective duty cycle was increased from 32 ± 10% to 52 ± 15% for a tumor motion with 4 s and 1.0 cm peak‐to‐peak. More reduction in the TIST was observed from 45.1 to 72.1% with an increase of the period from 2 to 8 s; effect of reduction was degraded by 54.5 to 46.2% when the peak‐to‐peak increased from 0.5 to 4.0 cm. The TIST increased when a field size formed by x‐jaws increased (correlation coefficient: 0.7). Conclusions: : Total treatment time was reduced noticeably with the algorithm presented in this study so that real‐time tumor tracking can be delivered with step‐and‐shoot IMRT with an increased duty cycle. This research was supported in part by a NIH grant 1R01CA133539‐01A2 and in part by the Intramural Research Program of the NIH, NCI.

SU‐FF‐T‐235: Patient‐Specific QA of Intensity‐Modulated Arc Therapy with 2D Diode Array: Initial Experience

Purpose: New rotational intensity‐modulated radiation therapy(IMRT) or intensity‐modulated arc therapy (IMAT) techniques using single or multiple arcs are attaining widespread adoption in clinic due to their superior delivery efficiency. This study evaluates the use of a 2D diode array for patient‐specific QA for such techniques with the Varian RapidArc™. Method and Materials: A RapidArc plan conforming to the clinical standards was generated for each case using Eclipse treatment planning system (Varian Medical Systems, Palo Alto, CA). A verification plan was subsequently created for each of the treatment plans with the water‐equivalent MapPHAN™ QA phantom, with the MapCHECK™ (Sun Nuclear Corporations, Melbourne, FL) embedded at 5 cm depth, measuring the coronal dose plane in integration mode. Twenty‐one cases were selected in this study including 2 brain, 11 head‐and‐neck (HN), 4 prostate, and 4 whole pelvis cases. The measured 2D dose distributions were then compared to that calculated by Eclipse using (i) gamma analysis with the acceptance criteria of 3%/3mm, and (ii) absolute point dose difference. Results: For all cases, the average passing rate for the gamma analysis was 98.2% (range: 95.3% – 100%) and the absolute point dose difference was 2.1% (range: 0%–3.9%). The total time taken for the QA process was approximately 10 – 15 minutes, including setup, plan delivery, and online analysis. Conclusions: Preliminary results have shown that a 2D diode array is an efficient method for RapidArc patient‐specific QA, proving that the MapCHECK/MapPHAN is capable of performing both absolute and relative dose comparisons with a satisfactory accuracy for clinical practices. Research sponsored by Varian Medical Systems.

MO-FF-A1-01: Fluoroscopic Verification of Intensity-Modulated Rotational Arc Therapy Delivery

Purpose: Intensity modulated rotational arc technique requires verification of leaf positions, gantry angle and dose rate in the entire arc. This study shows how to achieve this with a detailed verification of Varian RapidArc using a fluoroscopic electronic portal imaging device(EPID).Materials and Methods: Three Rapid Arc plans (prostate 1, whole pelvis 1, and head and neck 1) are delivered on a Triology linac (Varian Medical Systems, CA). During delivery, approximately 600 fluoroscopic portal images are acquired (∼8 images/s) per arc with a PV‐aS1000 EPID, without use of secondary phantoms or blocks. Each leaf position of each gantry angle is calculated from the acquired EPIDimages offline. Gantry angle information of each portal image is acquired from the dynalog file generated during beam delivery. Leaf positions from the dynalog file are compared to scheduled positions from the DICOM RT plan file. Results: Online EPIDimage acquisition of Rapid Arc delivery is prompt, involving extension of the EPID system and beam delivery time. The measurement error depends on the displacement of EPID system relatively to the center of rotation, which is only 1mm–1.5 mm. Offline analyses show the accuracy of leaf positions for static leaf and gantry field are better than 1 mm. More than 98.5% of leaf sequences exhibit less than 3mm deviations, 83 % show 2mm and 56% for 1mm. Conclusions: Position of each leaf of each gantry angle for Rapid Arc delivery is verified within 1 mm accuracy with fluoroscopic portal images. Use of fluoroscopic EPIDimages can be considered as a practical QA tool for the verification of the Rapid Arc delivery. This study is partially supported by Varian Medical Systems.

SU‐FF‐T‐182: Validation of Amplitude and Phase Gating Against Potential Irreproducibility of Motion

In gated radiation therapy, reproducibility of patients internal motion between imaging and delivery is essential. However, changes in period, amplitude, shape of motion, and baseline shift occur. In this study, we investigated the dosimetric impact of the changes on the coverage of clinical and internal target volumes for phase and amplitude gating. We used conventional and intensity‐modulated beams that are designed to cover internal target volumes. We assumed a duty cycle of 40‐to‐60% phases and equivalent amplitudes, a period of 4.5sec, and an amplitude of 4cm as conditions used in imaging. We introduced the above changes from these and used actual patients breathing motion in the measurement and computational simulation on a diode array (double precision). When a baseline shift of ‐1cm was assumed, only a part (67%) of clinical target volume (CTV) received a prescribed does for phase gating; a similar profile shift in delivereddose was observed for the amplitude gating (when the amplitude window was stationary). As the amount of the shift increased, the impact increased. An amplitude change by a few centimeters caused a shift in delivereddose profile and underdose in CTV for phase gating. The underdose was not observed for amplitude gating. The change in breathing periods did not affect the delivereddose profile. The change in breathing pattern from sinusoidal into linear shapes showed underdose in 12% of CTV and profile change for phase gating and no profile change and underdose for amplitude gating. The simulations agreed with the measurements, and used for the amplitude gating study. Results based on actual breathing patterns and intensity‐modulated fields will also be presented. This study has demonstrated that unless the amplitude opening is not adapting to the movement and extent of CTV, gated therapy is susceptible to the irreproducibility. Partly supported by Varian Medical Systems, Inc.

SU‐FF‐J‐132: A Novel Technique for Closed‐Loop Feedback Real‐Time Tracking by Controlling Dose Rate : A Parametric Study

Purpose: We have developed a new technique called dose‐rate‐modulated tracking (DRMT) for closed‐feedback real‐time tumor tracking by changing the dose rate. Method and Materials: DRMT uses pre‐programmed MLC sequences that are generated using measured data for tumor motion obtained at an earlier time. Since the leaves move, their positions are programmed with schedules that are a function of the accumulated dose. The leaf trajectory on the time axis then can be changed during treatment by changing the dose rate. DRMT changes the dose rate to minimize the discrepancy between the scheduled MLC position and the target position or breathing signal on the day of treatment. If the monitored breathing is slower than that observed at simulation, the dose rate is lowered to slow down the movement of the MLC and vice versa, thereby maintaining synchrony. DRMT tumor‐tracking power was tested with sinusoidalbreathing functions and patient‐breathing signals (RPM, Varian). The tracking error (2σ) for each breathing signal was derived as a function of the system reaction time and the dose‐rate correction period. Results and Conclusions: DRMT simulation showed that for the sinusoidal‐breathing signal with 2‐cm peak‐to‐peak tumor motion, less than 0.2 cm tracking error was obtained if the system can react to a detected mismatch in less than 0.3 s and the dose rate can be adjusted in 0.36 s. Fourteen out of 26 patients were eligible for DRMT. The selection criteria were: (1) peak‐to‐peak breathing motion greater than 0.5 cm and (2) amplitude variation less than 20 % during the DRMT simulation. The tracking error for the patient data is expressed as a percentage of the motion amplitude. A tracking error of less than 20 % was achieved for 13 out of the 14 patients if the system can respond in 0 s.

TH-EF-BRB-09: Total Body Irradiation with Uniform MU and Modulated Arc Segments, UMMS-TBI

PURPOSE To test a novel total body irradiation (TBI) system using conformal partial arc with patient lying on the stationary couch which is biologically equivalent to a moving couch TBI. This improves the scanning field TBI, which is previously presented. METHODS The Uniform MU Modulated arc Segments TBI or UMMS-TBI scans the treatment plane with a constant machine dose rate and a constant gantry rotation speed. A dynamic MLC pattern which moves while gantry rotates has been designed so that the treatment field moves same distance at the treatment plane per each gantry angle, while maintaining same treatment field size (34cm) at the plane. Dose across the plane varies due to the geometric differences including the distance from the source to a point of interest and the different attenuation from the slanted depth which changes the effective depth. Beam intensity is modulated to correct the dose variation across the plane by assigning the number of gantry angles inversely proportional to the uncorrected dose. RESULTS Measured dose and calculated dose matched within 1 % for central axis and 3% for off axis for various patient scenarios. Dose from different distance does not follow the inverse square relation as it is predicted from calculation. Dose uniformity better than 5% across 180 cm at 10cm depth is achieved by moving the gantry from -55 to +55 deg. Total treatment time for 2 Gy AP/PA fields is 40-50 minutes excluding patient set up time, at the machine dose rate of 200 MU/min. CONCLUSION This novel technique, yet accurate but easy to implement enables TBI treatment in a small treatment room with less program development preparation than other techniques. The VMAT function of treatment delivery is not required to modulate beams. One delivery pattern can be used for different patients by changing the monitor units.

SU-F-J-118: On-Treatment 4D CT Reconstruction From Planning 4D CT Using Linear Amplitude Scaling

PURPOSE To reconstruct on-treatment 4D CT images from planning 4D CT images by adapting deformation vector field (DVF) of the planning CT to the on-treatment condition, while the adaptation is based on the scaling of two amplitudes that are motion characteristics at the times of treatment and planning CT acquisition, respectively. METHODS An anthropomorphic digital phantom (XCAT) was used to generate 4D image sets with 1-cm and 2-cm tumor motions simulating conditions of planning CT and treatment, respectively. DVFs were acquired from the planning CT image set. The DRR images were acquired simulating setup kV images from the two CT image sets. On the DRR images, tumor positions and their motion amplitudes were quantified. The DVFs were scaled linearly by the amplitude ratio between the treatment and the planning CT times, assuming the elasticity of lung. The scaled DVFs were used to resample the planning 4D CT images generating on-treatment 4D CT images. The on-treatment 4D CT images thus acquired were compared with the reference on-treatment images (2-cm motion). RESULTS The resampled images showed good agreement within 1 mm residual errors with the reference images. The normalized cross correlation was 0.995. CONCLUSION A linear model of amplitude scaling was developed to reconstruct on-treatment 4D CT images from planning 4D CT images using the setup KV images acquired during treatment. The model was validated on a digital phantom. For the model to fully work, a further research needs to be followed, that aims at utilizing a phase-specific CT image set that is geometrically identical between pretreatment and treatment conditions.

SU-F-J-32: Do We Need KV Imaging During CBCT Based Patient Set-Up for Lung Radiation Therapy?

PURPOSE To evaluate the role of 2D kilovoltage (kV) imaging to complement cone beam CT (CBCT) imaging in a shift threshold based image guided radiation therapy (IGRT) strategy for conventional lung radiotherapy. METHODS A retrospective study was conducted by analyzing IGRT couch shift trends for 15 patients that received lung radiation therapy to evaluate the benefit of performing orthogonal kV imaging prior to CBCT imaging. Herein, a shift threshold based IGRT protocol was applied, which would mandate additional CBCT verification if the applied patient shifts exceeded 3 mm to avoid intraobserver variability in CBCT registration and to confirm table shifts. For each patient, two IGRT strategies: kV + CBCT and CBCT alone, were compared and the recorded patient shifts were categorized into whether additional CBCT acquisition would have been mandated or not. The effectiveness of either strategy was gauged by the likelihood of needing additional CBCT imaging for accurate patient set-up. RESULTS The use of CBCT alone was 6 times more likely to require an additional CBCT than KV+CBCT, for a 3 mm shift threshold (88% vs 14%). The likelihood of additional CBCT verification generally increased with lower shift thresholds, and was significantly lower when kV+CBCT was used (7% with 5 mm shift threshold, 36% with 2 mm threshold), than with CBCT alone (61% with 5 mm shift threshold, 97% with 2 mm threshold). With CBCT alone, treatment time increased by 2.2 min and dose increased by 1.9 cGy per fraction on average due to additional CBCT with a 3mm shift threshold. CONCLUSION The benefit of kV imaging to screen for gross misalignments led to more accurate CBCT based patient localization compared with using CBCT alone. The subsequently reduced need for additional CBCT verification will minimize treatment time and result in less overall patient imaging dose.