Proceedings of the 52nd Annual ASTRO Meeting S671

Purpose/Objective(s): To improve the accuracy of the calculated dose to the organs of the thorax during stereotactic body radi-ation therapy (SBRT) for non-small cell lung cancer (NSCLC), to better understand the relationship between radiation dose andorgan complications.

Materials/Methods: A cohort of 25 NSCLC patients treated under a 60 Gy in 3 fractions, 48 Gy in 4 fractions, or 54 Gy in 3fraction SBRT protocol, with complete 4D-CT and 4DCBCT imaging were retrospectively evaluated. The organs of the thoraxwere contoured on 4D-CT datasets obtained at treatment planning. MORFEUS, a finite element model-based deformable registra-tion algorithm with the capability to model the sliding interface between the lungs and the rib cage was used to model the breathingmotion. The model included trachea, central bronchus (up to 2nd bifurcation), distal bronchial tree, lungs, heart, esophagus, ribsand tumor. Radiation dose was calculated in a commercial treatment planning system on 4D-CT Inhale and Exhale images andaccumulated over the breathing cycle using MORFEUS. Dose results for the breathing model were compared to standard treatmentplanning (exhale images).

Results: Analysis has been completed on a subset of patients. The mean maximum dose deviation to a point between the breathingmodel and standard planning for the bronchial tree was 14 Gy (5 Gy-20 Gy). All patients showed dose deviations of more than 2 Gyor larger to 5 percent of the bronchial tree within the treated lobe of the ipsilateral lung. One patient showed dose changes of 5 Gy orlarger to 22 percent of the treated bronchial tree. Change in mean lung dose (range: 0-0.05 Gy) and lung V20 (range: 0.05%-1.1%)were insignificant. Similarly, changes in max dose and mean dose to the ribs, trachea, heart and esophagus were small overall andclinically insignificant (0.02 Gy-0.44 Gy). Dose deviations within the central bronchus were larger, with one patient experiencinga deviation of 2 Gy. Larger deviations in dose were also seen within the GTV for some patients, with a mean reduction in minimumdose to the tumor of 2 Gy (0.2 Gy-5.5 Gy) when breathing motion and deformation were modeled. Conversely one patientexperienced an increase in GTV dose of 8 Gy.

Conclusions: Results indicate that significant deviations from planned dosages can occur within the thorax when breathing anddeformation are considered. This is particularly true within the bronchial tree, GTV, and central bronchus. The use of patientspecific biomechanical models with the ability to account for patient breathing are an excellent tool for reducing the uncertaintyin determining the actual dose received by patients during treatment, and thus represent an important first step toward relatingradiation dose to organ complications in order to develop an NTCP model for lung SBRT.

Author Disclosure: S.F. Hunter, None; A. Al-Mayah, None; J. Moseley, None; A. Bezjak, Elekta Oncology Systems, C. OtherResearch Support; K.K. Brock, Phillips Medical Systems, B. Research Grant; Varian Medical Systems, B. Research Grant;Ray Search Laboratories, B. Research Grant; IMPAC advisory board member, F. Consultant/Advisory Board.

3046 Biofeedback during 4D-CT Image Acquisition Does Not Enhance the Reliability of ITVMIP for Radiation

Treatment Planning of Thoracic and Abdominal Malignancies: A Prospective Trial

G. A. Neuner, W. Lu, Z. Wang, S. Sassor, R. George, S. J. Feigenberg, W. F. Regine, W. D. D’Souza

University of Maryland Medical System Dept of Radiation Oncology, Baltimore, MD

Purpose/Objective(s): Although the contouring of ten-phase 4D-CT images resulting in the definition of an ITV10 can be cum-bersome, ITvs. generated from CT scans using Maximum Intensity Projection (MIP) reconstruction have been shown to underes-timate ITV10. We conducted a prospective trial to evaluate two biofeedback techniques hypothesized to improve ITV10/ITVMIP

correlation.

Materials/Methods: Patients enrolled had thoracic or abdominal tumors. At each session (3 sessions total), patients underwent: (1)an uncoached 4D-CT; (2) a 4D-CT scan while receiving audio and visual cues to maintain a regular breathing pattern; and (3) a 4D-CT while breathing into a spirometer, seeing a trace of their flow-volume loop, and receiving audio cues. Phase-based reconstructionwas utilized to generate 3D CT data sets corresponding to 10 phases of the respiratory cycle. ITV10 volumes were generated by com-bining the GTvs. from all 10 phases, and ITVMIP volumes were created from GTvs. drawn on the CTs generated via MIP reconstruc-tion. One physician (GAN) contoured all scans utilizing the same CT window level to generate both the ITV10 and ITVMIP. Volumedifference, centroid distance, Hausdorff distance, overlap index, and root mean square distance (RMS) between ITV10 and ITVMIP

were compared. Means were compared with independent sample t-tests.

Results: Eleven patients (6 with thoracic tumors, 5 with abdominal tumors) with 88 completed 4D-CTs were included in thisanalysis. The mean tumor volumes for all scans were 149 cc (4.6-539.6 cc) for ITV10 and 124 cc (4.2-456.1 cc) for ITVMIP (p\ 0.001). Comparing ITvs. from coached and uncoached 4D-CTs, there was no difference in absolute volumes or the ITV10/ITVMIP ratios. Comparing ITV10 and ITVMIP, the mean centroid distance was 1.7 mm (0-4.9 mm), the mean Hausdorff dis-tance was 10.7 mm (2.8-23.3 mm), the mean RMS distance was 2.7 mm (0.8-6.2 mm), and the mean overlap index was 0.78(0.56-0.92). When compared to parameters generated from uncoached 4D-CTs, neither biofeedback technique resulted in a sig-nificant difference in mean centroid distances (abdomen p = 0.66, thoracic p = 0.91), mean Hausdorff distances (abdomen p =0.82, thorax p = 0.97), mean RMS distances (abdomen p = 0.89, thorax = 0.87), or mean overlap indices (abdomen p = 0.87,thorax p = 0.81).

Conclusions: In this prospective cohort, mean ITVMIP is nearly 20% smaller than the mean ITV10 with all 4D-CT techniquestested. Biofeedback using the two methods described did not improve the match between ITVMIP and ITV10 in any parameterexamined in patients with both thoracic and abdominal tumors. Based on this, we recommend caution in the use of ITVMIP fortreatment planning regardless of 4D-CT technique if the goal is to fully account for tumor motion.

Author Disclosure: G.A. Neuner, None; W. Lu, None; Z. Wang, None; S. Sassor, None; R. George, None; S.J. Feigenberg, None;W.F. Regine, None; W.D. D’Souza, The Work in this Abstract was Funded by Philips Healthcare, B. Research Grant.

3047 Dosimetric Evaluation of Individualized Adaptive Motion Margins for Abdominal and Thoracic Tumors

P. R. Poulsen1, B. Cho2,3, P. J. Keall2