S736 I. J. Radiation Oncology d Biology d Physics Volume 78, Number 3, Supplement, 2010

3187 A Method to Estimate Three Dimensional Intrinsic Dosimetric Uncertainty Resulting from using

Deformable Image Registration for Dose Mapping

F. J. Salguero, N. K. Saleh-Sayah, C. Yan, J. V. Siebers

Virginia Commonwealth University, Richmond, VA

Purpose/Objective(s): To present a method to quantify dose uncertainties introduced by deformable image registration (DIR)based dose mapping processes.

Materials/Methods: The method estimates the DIRs spatial variance then applies it to the dose distribution to determine the dosevariance. To determine the DIR spatial variance, an iterative scheme is used to obtain a set of related images. In applying the DIRalgorithm to map a source image S to a target image T the result is a slightly different image T’ because of intrinsic inaccuracies. T’is mapped back to S, obtaining another image S’. In the next iteration, S’ is warped to T (resulting T’’) and T’’ is warped back not toS’, but to the original source image S. This assures that the images will not diverge after several iterations. Multiple iterations resultin a set of images S’, S’’ . whose points can be traced back to points in S. For each point of S, a cloud of points is obtained. Thecovariance matrix of the point displacement is estimated for every voxel. Dose SD is then estimated by the distribution of dosevalues around every point. Neighboring doses which could be mapped to the selected position are weighted by the estimated dis-placement uncertainty 3D Gaussian value at their position. A 74 Gy 4D lung treatment using Pinnacle3’s TPS DIR algorithm is usedto evaluate both displacement and dose uncertainty.

Results: Displacement and dose uncertainty maps were calculated for the test case. For this case, the maximum jSj1/2, with S beingthe covariance matrix, is 5.47 mm, whereas the maximum sdose is 21.81 Gy. The volume with sdose.10 Gy is 14.22 cm3. DVHscomputed incorporating dose uncertainties show that the target is minimally affected by displacement-uncertainty induced doseuncertainty, while lungs and heart DVH differences are greater.

Conclusions: The proposed method is an useful tool to find possible problematic regions for DIR algorithms and their dosimetricimpact in clinical treatments and provides useful metrics to evaluate the potential clinical significance of DIR inaccuracies.(Supported by NIH-P01-CA116602.)

Author Disclosure: F.J. Salguero, None; N.K. Saleh-Sayah, None; C. Yan, None; J.V. Siebers, Research contract with Philips Med-ical Systems, C. Other Research Support; Research contract with Varian Medical Systems, C. Other Research Support.

3188 Repeat ABC-breath Hold Imaging with Cone-beam CT

J. Boda-Heggemann1, F. Lohr1, H. Wertz1, I. Lob1, B. Kupper1, A. Kavanagh2, V. N. Hansen2, M. Brada2, F. Wenz1, H. McNair2

1Universitaetsmedizin Mannheim, Mannheim, Germany, 2Royal Marsden NHS Foundation Trust and Institute of CancerResearch, Radiotherapy Department, Surrey, United Kingdom

Purpose/Objective(s): Dose-escalated hypofractionated stereotactic radiotherapy (SBRT) for medically inoperable stage I-II lungcancer has shown a 3-year local control of 98% with minimal toxicity in the most recent update of RTOG-0236. Online image-guidance for pulmonal SBRT is a special challenge due to breathing-induced motion. For the daily clinical routine, breath-holdtechniques have been established for the reduction of respiratory motion. Target immobilization with ABC (Active Breathing Con-trol, Elekta, UK) has been shown to be an accurate and clinically useful tool in reducing the effect of breathing motion/blurring,making PTV margin-reduction and dose escalation possible. Cone-beam CT (CBCT) systems can acquire multiple radiographsduring a rotation of typically 200� to 360� with an acquisition time of 60-120s. Online 3D breath-hold treatment verification is,however, still an issue due to imaging times exceeding maximum breath hold times of ?20s for average patients. We describean imaging approach that can be performed in combination with ABC-based repeated breath hold with a commercially availableCBCT, making breath hold imaging and performing 3D-3D soft-tissue-based matching possible.

Materials/Methods: Treatment planning CT was acquired in moderately deep inspiratory ABC-based breath hold. CBCT volumescans were acquired before each fraction. Projection images were acquired using two methods: A full scan of rotation of 360� in 2min or a half scan of rotation of 200� in�70s. During one rotation, repeated breath hold phases were initiated for a predeterminedtime depending on individual patient characteristics, typically 15-20s, with intermittent short breaks of free breathing. Planning CTimages were matched with the CBCT images regarding soft tissue anatomy.

Results: Positioning based on ABC-based repeat breath hold has been applicable to .80% of our patients. The projections ac-quired during 3 half scans of 7 patients were analyzed and the mean (range) percentage of time spent in breath hold was 66%(52-81%). Repeated breath hold general image quality is comparable to the planning CT and better than free breathing CBCT.Some blurring of the image results from components of the image data acquired under motion but does not obscure the tumor/lung interface and direct 3D-3D soft tissue matching can be performed online, thus an universal workflow for all lung lesionscan be adapted.

Conclusions: ABC-based repeated breath hold imaging with a commercially available cone-beam-CT and performing 3D-3D soft-tissue-based matching is a feasible approach for image guided extracranial precision lung radiotherapy. Fast 3D imaging during onebreath hold, combining kV and MV imaging on a single linac is currently under development and has the potential to dramaticallyaccelerate clinical linac based volume imaging.

Author Disclosure: J. Boda-Heggemann, None; F. Lohr, None; H. Wertz, None; I. Lob, None; B. Kupper, None; A. Kavanagh,None; V.N. Hansen, None; M. Brada, None; F. Wenz, None; H. McNair, None.

3189 TomoTherapy’s Exit Detector Data for In vivo Quality Assurance of Head and Neck Treatments

S. Goddu, V. Rodriguez, J. Cates, O. Wooten, M. Michaletz-Lorenz, W. Thorstad, H. Gay, S. Mutic, G. Olivera, L. Daniel

Washington University, St. Louis, MO

Purpose/Objective(s): Helical Tomotherapy (HT) delivers conformal dose distribution in a complex manner and quality assur-ance of such treatments is essential for patient safety. Errors in patient positioning may result in either underdose or overdose