Proceedings of the 49th Annual ASTRO Meeting S703

breast MRI protocols were modified. Theoretically, voxel volume reduction would reduce the artifacts, and thus spatial resolutionalterations were made in the sequences within visually acceptable tolerance of SNR decrease. Echo time (TE) was decreased usingfast radiofrequency pulsing and asymmetric echoes to achieve the shortest TE possible. Extended echo trains were used whereacceptable to optimize refocusing effects. This is also why 3D techniques were used over 2D techniques when applicable or accept-able. Receiver and transmitter bandwidth were increased as much as possible to minimize their effects in susceptibility artifacts.Frequency selective fat suppression was not used. The implant distortion volume varied between fast spin echo and gradient echosequences in an anticipated fashion. Image distortion was improved by the sequence alterations utilized, but was not eliminated.Further testing of modified sequences is warranted for further optimization.

Conclusions: Recent literature has shown has shown that frequency selective fat suppression exacerbates the artifacts created bymetallic implants in models testing breast MRI imaging protocols (2, 3). We did not use frequency selective fat suppression in ourprotocol, favoring non-suppressed T1 images with computer aided volume subtraction post-processing instead. This imaging tech-nique, in combination with sequence parameter changes described above, allowed further reduction in the image artifact caused bythe DVS on MRI.

1. Int J Rad Onco Bio Phys 2005;62:606–6132. AJR 1999;172:1417–14193. AJR 2007;188:372–376

Author Disclosure: C.W. Scarantino, Shareholder, E. Ownership Interest; Medical Director, F. Consultant/Advisory Board; G.P.Beyer, Medical Physicist, A. Employment; G.G. Coates, None; C. Rini, Engineering, Product Development, A. Employment.

2897 Temporal Delivery Efficiency of a Novel Single Gantry Arc Optimization Technique for Treatment of

Recurrent Nasopharynx Cancer

K. Otto, M. Milette, J. Wu

British Columbia Cancer Agency, Vancouver, BC, Canada

Purpose/Objective(s): The adoption of on-line imaging technologies will inherently increase overall treatment times. Further-more, the benefits of daily repositioning or plan adaptation will be limited by patient motion or deformation that occurs duringthe delivery. The purpose of this work is to evaluate the time efficiency of a novel technique where treatment delivery is performedin a single gantry arc.

Materials/Methods: An optimization platform for planning and delivering radiation therapy in a single gantry arc is used to gen-erate plans for 10 recurrent nasopharynx cancer (rNPC) patients. PTVs vary in size, shape and proximity to healthy tissue struc-tures. Due to previous radiation treatment strong dose-volume constraints are used to limit dose to critical structures. Delivery isperformed in a single arc where the gantry rotates continuously through 228 degrees biased symmetrically about the anterior aspect.MLC leaf positions are optimized directly and change position continuously throughout the rotation. For comparison, patients werealso planned using a static gantry position IMRT technique using a commercial planning system (Eclipse 6.5-Varian Medical Sys-tems). Dose distributions for the single arc technique were generated so that DVHs were equivalent to or better than the static gantrymethod. The static gantry plan consists of 7 beams spread uniformly throughout the same gantry range used for the single arc. De-livery is performed using the dynamic sliding window technique which is currently considered to be the most time efficient IMRTdelivery method. The single arc and static gantry plans were delivered on a Varian CL21EX linac using a dose rate of 300 MU/minute. Single arc treatments may be delivered with either constant or variable Monitor Units (MU) per degree of rotation. Deliverytime is defined as the time between the first and the last MU for a treatment fraction. Time for patient set-up, imaging, repositioningor adaptation is not included.

Results: For the patients evaluated so far the mean treatment time for the single arc technique with variable and constant MU perdegree was 2.4 and 1.7 minutes respectively. It was noted that forcing the constant MU per degree constraint resulted in some de-terioration of plan quality compared to the other techniques. The static gantry sliding window IMRT plans were delivered in a meantime of 6.4 minutes. It should be noted that there is a delay between each static IMRT beam due to data transfer between the treat-ment computer and the patient database. Because a single arc is considered an individual beam at the treatment console there is nodelay for this method. The mean treatment time for static gantry delivery would be reduced to 4.4 minutes if this overhead could beeliminated.

Conclusions: For rNPC the novel single arc technique offers a substantial reduction in treatment time over the static gantry slidingwindow technique. Shorter delivery times should result in less geometric inaccuracy making single arc delivery well suited foradaptive radiation therapy.

Author Disclosure: K. Otto, Varian Medical Systems Inc., B. Research Grant; Varian Medical Systems Inc., F. Consultant/Advi-sory Board; M. Milette, None; J. Wu, None.

2898 Navigating Through and Around the Field Size Limitation on Varian Linear Accelerators When Planning

Large IMRT Cases Using the Philips Pinnacle Treatment Planning System

V. L. LaCerba1, K. L. Simon2, B. E. Nelms3, E. G. Hendee4

1Radiation Oncology Resources, McMinnville, OR, 2Radiation Oncology Services, Inc., Riverdale, GA, 3Canis Lupus, LLC,Madison, WI, 4Waukesha Memorial Hospital, Waukesha, WI

Purpose/Objective(s): To determine the best method by which the result of the field width limitation on the Varian Linear Ac-celerator is minimized in the final treatment plan. There is a maximum field width on the Varian linear accelerators that is oftensmaller than the width of some PTV’s in large IMRT cases. This often presents a difficult planning problem that must be navigatedaround during the planning process in order to achieve the best clinical result and outcome.

Materials/Methods: A total of 5 head and neck patients and 5 pelvis patients were planned using 4 different treatment planningtechniques. These patients were planned using split field, wide field, limited field, and with solid IMRT compensators. These were

S704 I. J. Radiation Oncology d Biology d Physics Volume 69, Number 3, Supplement, 2007

all planned within the Philips Pinnacle Treatment Planning System using version 7.4. Also plans were also planned using directmachine parameter optimization (DMPO). All plan prescriptions were prescribed using a mean dose to the highest dose target vol-ume. Isodose lines were chosen based on the isodose line which was able to deliver at least 95% coverage to each of the PTV’swithin the treatment plans.

Results: Each of the different planning techniques showed advantages and disadvantages. The split field technique allowed thetumor to be completely covered in each beam angle but also ended up having the highest number of MU’s. This increased the pa-tient’s time on the table greatly. When planned with the limited field technique, monitor units were decreased but the tumor was notfully covered by every beam. In some cases, addition beams were needed and added in order to obtain adequate coverage. The widefield technique allowed the volumes to be covered fully in every beam but there was an increased amount of leakage where theMLC’s closed within the field. This is of concern when the junctions occurred over critical structures (ie: spinal cord, brainstem,and chiasm). The last technique compared was solid IMRT. These plans were also planned using decimal solid compensators. Thesolid IMRT plan provided the most uniform tumor dose coverage but some of the critical structures received higher doses due to theminimum transmission through the compensators.

Conclusions: The wide and limited field techniques provided adequate coverage while lowering the total monitor units of the plan.The technique that seemed to be most undesirable, while most widely used within the therapy community, was the split field tech-nique. Of all of the techniques the solid compensators provided the most uniform dose distribution to our PTV volumes. The solidcompensator plan also had the fewest total of monitor units of all of the planning techniques. This resulted in the least amount oftime on the treatment table for the patient.

Author Disclosure: V.L. LaCerba, None; K.L. Simon, None; B.E. Nelms, None; E.G. Hendee, None.

2899 Radiotherapy With a Flattening Filter Free Clinac

O. N. Vassiliev, S. F. Kry, R. Mohan, U. Titt

University of Texas MD Anderson Cancer Center, Houston, TX

Purpose/Objective(s): To assess the potential of photon radiotherapy with a treatment delivery system based on a Varian Clinacwith the flattening filter removed from the beam-line.

Materials/Methods: Comprehensive measurements were taken on a Varian Clinac 21EX (Varian Medical Systems) operatedwithout a flattening filter, in 6 and 18 MV photon modes. Results of the measurements were complemented with Monte Carlo sim-ulations and extended to 8, 10, and 15 MV photon beams. Extensive data analysis was performed. Differences between flattenedand unflattened beams were examined in detail. With the acquired data a treatment planning system Eclipse 8.0 (Varian MedicalSystems) was commissioned. Treatment plans, including IMRT and stereotactic radiotherapy, for several treatment sites weredeveloped and analyzed.

Results: Results of the treatment planning studies support the feasibility of radiotherapy with the flattening filter free Clinac. Inmost cases studied it was possible to develop clinically acceptable treatment plans with unflattened beams, following planning tech-niques and templates normally used with flattened beams. In terms of patient dose distributions plans with unflattened beams weresimilar to plans with flattened beams, except close to the skin where unflattened beams deliver higher dose than flattened beams ofthe same energy. The skin dose, however, can be reduced by increasing energy of unflattened beams, for example, from 6 MV to 8or 10 MV. The main result of the removal of the flattening filter is a considerable increase in the dose rate and a respective decreasein beam-on time. For the treatment plans with 6 MV beams the typical time decrease was greater than a factor of 2, and much moreat 18 MV. Faster dose delivery can benefit the development of breath-hold and respiratory-gated treatment techniques. Anotheradvantage of the flattening filter free Clinac is the reduction in the amount of leakage and scattered radiation reaching the patientduring the course of the treatment. Dose to the patient from neutrons produced in high energy beams is also reduced. The statedreductions were confirmed through measurements and Monte Carlo simulations. These findings indicate potential reduction in thehealth risks associated with out-of-field radiation, secondary malignancies in particular.

Conclusions: The reviewed data support the feasibility of radiotherapy with a flattening filter free Clinac. This proposed treatmentdelivery system will achieve substantially higher dose rates and will lower doses outside the treatment field, as compared to theconventional Clinac.

Acknowledgements: The study was supported by Varian Medical Systems.

Author Disclosure: O.N. Vassiliev, Varian Medical Systems, B. Research Grant; S.F. Kry, None; R. Mohan, None; U. Titt, None.

2900 Correcting for Intra-Fraction Breathing Motion Based on Offline and Online Analysis of Respiratory

Patterns

J. Xu1, C. Shi2, S. B. Jiang3,4, N. Papanikolaou2

1University of Texas Health Science Center at San Antonio, San Antonio, TX, 2Cancer Therapy and Research Center, SanAntonio, TX, 3University of California, San Diego, San Diego, CA, 4Moores Cancer Center, San Diego, CA

Purpose/Objective(s): To develop an offline and online predictive model by analyzing the patient’s breathing pattern in order tocorrect for the intra-fraction motion effect during radiation therapy treatment.

Materials/Methods: A model based analysis of the breathing signal was performed to help predict motion associated with respi-ration. Two adaptive filters, the Kalman filter and linear predictive filter were implemented to predict the online breathing signal.The patient breath signal was acquired using a Varian Real-time Position Management (RPM) with a sample frequency 30 Hz. Toremove the noise from the signal, the first-order smoothing filter was used. For the offline analysis, the piecewise cosine functionmodel was used as basic model using a partitioned nonlinear least squares algorithm to perform the model based analysis. For on-line prediction, the time lag was set to 200 ms. Both the Kalman filter and Linear Predictive filter were applied to evaluate themodel’s ability to predict the breathing signal.

Results: Figure 1 shows the patient’s breathing curve as well as the offline and online analysis results. The model based offlineanalysis can predict several parameters, including the expiration duration, the inspiration duration and amplitude of the breathingcycle. The model follows closely the breathing signal, except for the peak inspiration and expiration. The R-Square value of theprediction result with the Kalman filter is up to 0.95, while with the Linear prediction is 0.98.