S672 I. J. Radiation Oncology d Biology d Physics Volume 75, Number 3, Supplement, 2009

cm and 0.44 ± 0.57 cm for the thorax and abdomen patients respectively (p \ .0001). Five percent of thoracic fractions had fluo-roscopic shifts greater than 3 mm, while 52% of abdominal fractions had shifts greater than 3 mm (p\ .0001) The added time forfluoroscopic imaging was an average of 10 minutes after CBCT imaging.

Conclusions: Results suggest that for thorax patients additional couch shifts based on fluoroscopic imaging may not be needed,except for adjustment of the gating window. For accurate setup of abdominal cancer patients however, there is a definite need forshifts based on fluoroscopy of fiducial markers. This may be due to poor visibility of abdominal tumors, imaging artifacts from themetal fiducials in CBCT images or increased respiratory motion of abdominal tumors compared to lung tumors.

Author Disclosure: L. Santanam, None; J. Bradley, None; C. Noel, None; J. Esthappan, None; I. ElNaqa, None; P. Parikh, None.

3093 Comparing Static vs. Rotational IMRT for Spine Body Radiotherapy

Q. Wu, J. Kirkpatrick, S. Yoo, R. McMahon, D. Thongphiew, F. Yin, F. Yin

Duke University Medical Center, Durham, NC

Purpose/Objective(s): Volumetric arc modulated treatment (VMAT) reduces treatment time over static beam intensity modulatedtreatment (IMRT) by delivering IMRT using arcs. This clinical study evaluates the feasibility of using VMAT for spine stereotacticbody radiotherapy (SBRT) to achieve highly conformal dose distributions that spare adjacent OARs (organ-at-risk) with reducedtreatment time.

Materials/Methods: Ten spine SBRT patients were studied retrospectively. IMRT plans were generated using 8-12 fields, andVMAT plans were generated with either one or two arcs on the Eclipse planning system (Varian medical Systems). PTV dose cov-erage, OAR dose sparing (including cord, lung, esophagus, liver, heart and kidney), and normal tissue integral dose were measuredand compared. Differences in treatment delivery were also analyzed by comparing MLC segments, total MUs, and total treatmenttime. The non-parametric Wilcoxon signed rank test was used for statistical analysis with significance p \ 0.05.

Results: The PTV DVHs were comparable between the VMAT and the IMRT plans in the shoulder (D99%-D90%), slope (D90%-D10%), and tail (D10%-D1%) regions (p values: 0.064-0.686). When averaged over all plans, only VMAT2arc had a better conformityindex than IMRT (1.09 vs.1.15, p = 0.007). In terms of cord sparing, IMRT was the best and VMAT1arc was the worst. IMRTachieved .10% more D1% cord sparing for 6 out of 10 cases and 7-15% more D10% sparing over the VMAT1arc. There is a cor-relation between the total number of segments and the dose drop-off rate within the cord. More segments in the IMRT plans allowedfor finer intensity map manipulation, providing a steeper dose gradient at the PTV-cord junction in the IMRT plan. And dose spar-ing to the cord is improved by using 2-arcs. The difference between IMRT and VAMT2arc were smaller and statistically insignif-icant at all dose levels. The differences were small and statistically insignificant for other OAR sparing. The mean MUs were 8711,7730 and 6317 for IMRT, VMAT1arc and VMAT2arc plans, respectively, with a 26% reduction from IMRT to VMAT2arc. The meantreatment time was 15.86 minutes for IMRT, 8.56 for VMAT1arc and 7.88 for VMAT2arc. Interestingly, adding another arc did notincrease the treatment time. More MLC segments in the 2-arc technique reduced the total MUs over 1-arc, indicating that the in-creased freedom of MLC segments allowed for more efficient coverage of different PTV-OAR overlap regions. The average in-tegral dose was the lowest for IMRT and the highest for VMAT1arc. However, the difference was statistically insignificant.

Conclusions: Although VMAT provided comparable PTV coverage and esophagus sparing for spine SBRT, 1arc showed signif-icantly worse spinal cord sparing than IMRT, while 2arc was comparable to IMRT. Treatment efficiency is substantially improvedwith the VMAT technique.

Author Disclosure: Q. Wu, None; J. Kirkpatrick, None; S. Yoo, None; R. McMahon, None; D. Thongphiew, None; F. Yin, None;F. Yin, None.

3094 Single Segment Non-coplanar Beam Optimization for Gated Lung SBRT Planning and Delivery

J. Partouche, T. Wu, K. Farrey, J. K. Salama, K. M. Yenice

The University of Chicago Medical Center, Chicago, IL

Purpose/Objective(s): To investigate the potential benefits of non-coplanar large segmented direct machine parameter optimiza-tion (DMPO) for primary and metastatic lung SBRT. To assess the delivery accuracy of DMPO beam segments to a gated target ina motion phantom.

Materials/Methods: Seven patients with primary or metastatic lung lesions treated with SBRT according to our institutional IRBprotocol were included in this study. Patients with oligometastases received doses of 3 x 10-14 Gy and primary NSCLC patientsreceived 50-60 Gy in 3-10 fractions depending on the tumor size and location. 3D custom treatment plans used 10-12 non-coplanar6MV beam arrangements with manually optimized beam MLC aperture and weight to meet the clinical goals. Inverse planning wasdone using the identical beam arrangements. Beam apertures were reset and optimized through the DMPO (Pinnacle, Philips) al-gorithm with one segment per beam and the minimum segment area of half of the maximum PTV cross-section in the BEV. DMPOplans were normalized at the same identical target coverage level as those for 3D plans. The optimization routine used the col-lapsed-cone convolution dose calculation after 10 successive iterations. 3D dose distributions, dose volume histograms, and normaltissue complication probabilities (NTCP) were calculated for both 3D conformal and 1-segment/DMPO SBRT plans. Finally, a mo-tion phantom with a lung-equivalent insert was fitted with a small tissue-equivalent material to represent a tumor in lung. Gafchro-mic EBT films were fitted between the sections of this phantom and were irradiated with optimized single-segment beams toevaluate the dose in the lung, in the target and at the edge of the target.

Results: One-segment/DMPO planning improved the conformality of SBRT over 3D CRT delivering on average 12% ± 7%,8% ± 7%, and 2% ± 2% less dose to the lung at 20, 13, and 5Gy levels, respectively. The maximum dose to the heart, esophagus,and cord were comparable between the 1-segment DMPO and 3DCRT plans. Except for one case, there was no increase in thenumber of monitor units used with 1-segment/DMPO plans. One -segment/DMPO plans yielded a mean reduction of 17.1%and 30% in the normal lung EUD and estimated lung complication probability, respectively, over 3DCRT plans. A dose to distanceagreement of 3%/3 mm between calculation and film measurement for a representative plan in a motion phantom with gating wasverified at 99% of points within the fields.

Proceedings of the 51st Annual ASTRO Meeting S673

Conclusions: Single segment beam DMPO can be used to improve SBRT planning for lung lesions to meet the planninggoals in an effective manner. This approach allows for easily deliverable and verifiable beam apertures for gated beam de-livery. The automation of our method is a good alternative to more traditional methods and offers significant dosimetricbenefits.

Author Disclosure: J. Partouche, None; T. Wu, None; K. Farrey, None; J.K. Salama, None; K.M. Yenice, None.

3095 Dosimetric Impact of Angular Deviations in Positioning for Spinal Radiosurgery

J. Kim, J. Jin, N. Walls, N. Wen, D. Liu, S. H. Patel, B. Movsas, J. Rock, S. Ryu, I. J. Chetty

Henry Ford Health System, Detroit, MI

Purpose/Objective(s): Daily patient setup accuracy in external beam radiation therapy has improved significantly as a result ofadvances in image-based guidance systems. Most of these systems account for variations in both target translations and rotations.However, it is common practice at many centers to account only for translational shifts while leaving rotational errors uncorrected.The purpose of this study is to investigate the dosimetric impact of angular deviations in spinal radiosurgery.

Materials/Methods: Seven spinal lesions encompassing 1-2 vertebral sections in various locations from cervical to lumbar spinewere investigated. For each treatment isocenter, 18 artificially rotated CT images were generated from the simulation CT in therange of ±5 degrees around all three axes. The original treatment plans were applied to the rotated images, the correspondingdose matrixes were recalculated, and the dose volume histograms (DVH) of the planning target volumes (PTV) and spinal cordswere analyzed. The values of the PTV minimum dose (Dmin), PTV coverage (at prescription dose), maximum cord dose (Dmax), andthe cord dose at 10% volume (D10%) were compared with those of the original plans.

Results: The original PTV coverage was 95±2%. The PTV Dmin, cord Dmax and D10% were respectively 70 ± 5%, 73 ± 8%, and53 ± 6% of the prescription dose. The PTV coverage and Dmin decreased with increasing rotation angle, while the cord Dmax andD10% increased with the change in angle. For all rotated plans, the differences in PTV coverage and cord D10% from the originalplan were relatively small. The ratios of the mean PTV coverage of images rotated by 5 deg to the original no-rotation cases (5/0 degratio) were 0.99 ± 0.01, 0.98 ± 0.02, and 0.97 ± 0.01 around the L/R, A/P, and S/I axes respectively. The corresponding cord D10%ratios were 1.03 ± 0.06, 1.02 ± 0.02, and 1.06 ± 0.05. On the other hand, the differences of the PTV Dmin and cord Dmax wererelatively large with 5/0 deg ratios of 0.85 ± 0.13, 0.92 ± 0.10 and 0.87 ± 0.09 for PTV Dmin, and 1.16 ± 0.13, 1.07 ± 0.05,1.14 ± 0.07 for cord Dmax. In other words, the target cold spot increased from the induced rotations because a portion of it movedout of the beam path, and the cord hot spot increased because it moved closer to the high dose gradient area. These dosimetricdifferences due to 5-deg rotational variations were found to be comparable to those of 0.5-1 mm translations in the A/P axisand 1-2 mm translations in the other two axes.

Conclusions: The dosimetric impact of angular deviations in target positioning for spinal radiosurgery is relatively small, espe-cially for tumors with smaller dimensions in the longitudinal direction.

Author Disclosure: J. Kim, None; J. Jin, None; N. Walls, None; N. Wen, None; D. Liu, None; S.H. Patel, None; B. Movsas, None;J. Rock, None; S. Ryu, None; I.J. Chetty, None.

3096 Online Monitoring of Body Stereotactic Treatments with Orthogonal kV-MV Imaging

R. C. Susil, E. Tryggestad, E. Ford, T. McNutt, J. M. Herman, J. Wong

Johns Hopkins University, Baltimore, MD

Purpose/Objective(s): While conebeam CT (CBCT) allows for accurate daily setup, it does not ensure that the setup remains validthrough the whole treatment session. This is of particular importance during body stereotactic treatments where small margins areemployed. We have developed and implemented a system for monitoring fiducials (e.g., gold seeds) and structures (e.g., bony anat-omy or implanted hardware) during radiotherapy treatment sessions with orthogonal kilovoltage (kV) and megavoltage (MV) pro-jection imaging. Here, we present our initial clinical experience with stereotactic body treatments.

Materials/Methods: Image visualization, processing, and interface routines were developed (MATLAB, Mathworks Inc.) andexecuted during treatment on a PC workstation adjacent to the Synergy S console (Elekta Inc.). Programmatic access to the Synergydatabase allows for online update and display of kV and MV projection images as they are acquired during treatment. Regions ofinterest and points of interest (imported from the Pinnacle3 planning system) are displayed as overlays on kV and MV images -allowing for rapid interpretation of structure and fiducial alignment during and between treatment fields.

Results: In our initial clinical work, this technique and system has been applied to support two stereotactic liver treatments, onestereotactic spine treatment, and one pancreatic treatment. For spine treatments, bony anatomy and surgical hardware aided in vi-sualization. All other patients had one or more 1x5 mm gold fiducial marker implanted. Daily patient setup was performed usingCBCT alignment. Subsequently, during and between treatment fields, kV and/or MV projection imaging was performed to visu-alize structures and/or fiducials within the treatment volume. Implanted gold markers were clearly visible on all kV and most MVprojection images. Fiducial misalignment (down to 1 mm) was easily and rapidly appreciated. When fiducials were identified out-side of the prescribed PTV margin, shifts were applied before continuing with treatment. Also, as we are able to acquire MV and kVprojections simultaneous with beam delivery, motion during a breath hold (in one case, due to patient discomfort) was detected andtreatment halted.

Conclusions: Using this technique and system, shifts that occur after setup, during treatment, can be quickly detected. Continuingclinical applications include spinal, thoracic, and abdominal radiosurgery (supporting two open, actively accruing clinical proto-cols). This system requires no hardware or software modification of the clinically available Synergy-S system and is also readilyadaptable to intensity-modulated arc therapy treatments. Continued work is focusing on feature detection algorithms for automaticshift determination.

Author Disclosure: R.C. Susil, None; E. Tryggestad, None; E. Ford, None; T. McNutt, None; J.M. Herman, None; J. Wong, None.