2844 Aperture-Based Modulated Electron Radiation Therapy (MERT)—An Efficient Approach to the Deliveryof MERT Plans

Y. Song1, S. Wang2, M. Chan1, A. Dhawan2, C. Burman1

1Memorial Sloan-Kettering Cancer Center, New York, NY, 2New Jersey Institute of Technology, Newark, NJ

Purpose/Objective(s): Currently, the most popular inverse treatment planning techniques fall into two distinct categories: 1)beamlet-based inverse treatment planning and 2) aperture-based inverse treatment planning. There are several problemsassociated with beamlet-based approach. First of all, optimized intensity maps are often converted into a series of discrete levelsin an attempt to make the leaf sequencing easier. However, this step also introduces some quantization errors and thereforeresults in loss of treatment plan quality. Secondly, leaf sequencing is constrained by hardware-related factors and therefore oftenrequires a large number of complex field shapes to deliver a given intensity map, thereby decreasing the overall deliveryefficiency. Thirdly, since the leaf sequencing step is excluded from the intensity map optimization process, therefore, all thedelivery-related effects, such as leakage, the tongue-and-groove effect, and head scatter are not taken into account whenoptimizing an intensity map. Aperture-based optimization is designed to reduce the complexity of treatment plans and is flexibleenough to easily include delivery-related effects. In particular, aperture-based optimization requires no leaf sequencing andapertures are guaranteed to satisfy delivery constraints. For electron therapy, plans can be delivered using electron cutouts ratherthan an electron-specific MLC (EMLC), which is not available on the current medical linear accelerators. It is therefore the mostpromising and practical approach to the delivery of modulated electron radiation therapy (MERT) plans.

Materials/Methods: In this study, we consider a new aperture-based inverse treatment planning technique for electron beamsthat optimize aperture shapes, weighting factors, energy selection, and bolus geometry if energy modulation does not producethe desired depth dose conformity. In this approach, the objective function is expressed in terms of the weight of each field.In other words, the weights of all the beamlets are equal within each field. This simplification remains throughout theoptimization process.

Results: Thus, the gradient direction along which the line minimization is carried out imposes the same changes in weights forall the beamlets that belong to the same field. Based on these conditions, the objective function is the sum of term that representsthe target prescription dose and other terms that represent violations to all the constraints that apply. The inverse treatmentplanning begins with the manual selection of beam angles that are enface to the skin. For each beam angle, a target depthhistogram is computed. The depth with highest beamlet frequency determines the electron energy for this field. The beamapertures are initially chosen to be the Beam Eye’s View of the target and the critical structures.

Conclusions: Our hope is naturally that optimal aperture shapes are somewhat close to the BEV and that making this choicewill save a significant amount of computation time. The plan is optimized over all the possible aperture shapes and weights.An optimal set of beam aperture shapes and weights is the one that minimizes the objective function. Once an optimal set ofbeam aperture shapes and weights are obtained, the plan can be further fine tuned by optimizing the bolus geometry andthickness. The final plan can be delivered with electron cutouts.

Author Disclosure: Y. Song, None; S. Wang, None; M. Chan, None; A. Dhawan, None; C. Burman, None.

2845 An Evaluation of IMRT Techniques for the Treatment of Para-Spinal Targets

B. H. Robison, C. R. Ramsey, R. M. Seibert

Thompson Cancer Survival Center, Knoxville, TN

Purpose/Objective(s): The goals of this study were to 1.) Compare IMRT treatment techniques for a para-spinal mass locatedin the thoracic spine of an anthropomorphic phantom, and 2.) Measure the accuracy of Megavoltage-CT (MVCT) images forlocalizing spinal anatomy in the T-Spine region.

Materials/Methods: Treatment planning CT images were acquired of a whole body anthropomorphic phantom. The treatmentplanning CT images of the phantom were acquired with a 1-mm slice thickness and a 50-cm field-of-view, which yielded aresolution of 0.98 x 0.98 x 1.00-mm. The planning target volume (PTV) was a geometrically constructed half annulus 25-mmthick in the anterior-posterior direction, and 15-mm thick laterally. A 10-mm in diameter spinal cord organ at risk (OAR) wascreated based on the center of the vertebral foramen. Fixed gantry IMRT and helical tomotherapy treatments were used toevaluate dose gradients and dose uniformity. Image fusion error was also measured by MVCT imaging the phantom’s T9vertebra multiple times without moving the phantom. A Principal Components Analysis (PCA) was then used to determine whatfactors most affected the image fusion.

Results: For sliding window plans, the calculated maximum OAR dose was 80% of the prescribed PTV dose for a 9-fielddelivery and 76% for a 12-field delivery. The calculated maximum OAR dose was 71% for a Step-and-Shoot 9-Field and 65%for a 12-Field. The measured dose gradient using film was 10% / mm for the tomotherapy delivery, and 14% / mm for 12-FieldStep-and-Shoot. The measured tomotherapy imaging system errors (1) were �1.3 mm, �0.8 mm, and �0.6 mm with theautomatic image fusion options set to “Bone”, “Bone and Soft Tissue”, and “Full-Image”. PCA showed that the “Bone” and“Bone and Soft Tissue” techniques had significant outliers and should not be used without careful manual review. Manualregistration after auto-fusion yielded the best agreement: (1) of �0.5 to �0.7 mm. PCA indicated that there was a strongrelationship between the full-image technique and manual registration. This result can most likely be attributed to performingthe manual registration directly after using the full-image technique.

Conclusions: Multiple techniques have been evaluated for the treatment of para-spinal targets using image-guided helicaltomotherapy and fixed-gantry based IMRT. Helical tomotherapy and 9-Field DMPO treatments yielded similar dose gradients(10%/mm) and PTV dose uniformity indices (10%). Anthropomorphic phantom studies indicated that megavoltage CT imageswere capable of imaging the spine with sufficient accuracy to place the isocenter within 1-mm of the desired position.

S686 I. J. Radiation Oncology ● Biology ● Physics Volume 66, Number 3, Supplement, 2006

Author Disclosure: B.H. Robison, None; C.R. Ramsey, TomoTherapy, Inc., D. Speakers Bureau/Honoraria; TomoTherapy,Inc., F. Consultant/Advisory Board; R.M. Seibert, None.

2846 A Dosimetric Study of Dynamic Conformal Arc Therapy for Lung Cancer

N. K. Cho, P. G. Maxim, B. W. Loo

Stanford Cancer Center, Stanford, CA

Purpose/Objective(s): IMRT is a potential means of escalating doses to non-small cell lung tumors while reducing doses tonormal tissues, particularly the uninvolved lungs. Despite dosimetric advantages over conventional 3-D conformal techniques,IMRT plans are time-consuming to deliver. Dynamic conformal arc therapy (DART) can also produce highly conformal dosedistributions, with time-efficient delivery. We performed a dosimetric comparison of IMRT and DART plans for a represen-tative case of lung cancer.

Materials/Methods: We compared IMRT and DART plans for a patient with stage IIIA (T2N2M0) non-small cell lung cancerinvolving the right lower lung and a right pretracheal lymph node. The targets were defined using PET-CT imaging, as wellas 4-D CT to determine margin expansion for organ motion. No elective nodal regions were targeted. Heterogeneity correctionwas used for all dose calculations, and all plans were normalized to achieve coverage of 95% of the target by the prescriptiondose of 70 Gy. DART plans were constructed as multiple evenly weighted coplanar fields at 10 degree increments of gantryrotation, each shaped by the MLC to conform automatically to the targets with a 5 mm margin for penumbra. There was noblocking of avoidance structures or inverse planning. We evaluated two DART plans: a single isocenter plan treating bothlesions in a 360 degree arc, and a two isocenter plan treating each lesion with a partial arc eliminating the longest paths enteringthrough normal lung. These were compared to six-field inverse-planned IMRT. We also compared each of the plans using 6and 15 MV beam energies.

Results: DVHs of the three plans are shown in the figure. PTV coverage was equivalent between the plans because of thenormalization. However, both DART plans had more homogeneous coverage with lower maximum doses. With respect tonormal lung doses, the single isocenter, full arc plan was marginally inferior to the IMRT plan in the �40 Gy range, whereasthe two isocenter, partial arc plan was superior at all dose levels. Both DART plans would be expected to result in substantiallyfaster delivery at one minute per revolution. For IMRT, 6 MV energy resulted in higher soft tissue hot spots outside the targetcompared with 15 MV, while beam energy had a negligible impact on the dose distributions of the DART plans.

Conclusions: For appropriately chosen targets, DART can result in better target dose homogeneity, comparable or better normaltissue dose distributions, and faster treatment time compared to IMRT. In addition, because equivalent plans can be generatedwith lower beam energy in DART, dose build-up effects in lung tumors may be less of a concern.

Author Disclosure: N.K. Cho, None; P.G. Maxim, None; B.W. Loo, None.

2847 Treatment Outcome-based Objective Functions for IMRT Treatment Planning

I. El Naqa, V. H. Clark, Y. Chen, M. Vicic, D. Khullar, S. Shimpi, A. Hope, J. Bradley, J. O. Deasy

Washington University, St. Louis, MO

Purpose/Objective(s): To investigate mathematical functions that may correlate with treatment outcome, but which are alsomore computationally efficient and radiobiologically reasonable than dose-volume constraints for IMRT optimization.

S687Proceedings of the 48th Annual ASTRO Meeting