- Email: [email protected]
Image Guided Stereotactic Radiosurgery (SRS) Treatment of Multiple Brain Metastases using Volumetric Modulated Arc Therapy (VMAT)
I. J. Radiation Oncology d Biology d Physics
S692
Volume 75, Number 3, Supplement, 2009
displacement relative to the LCF. The optical tracking system (OTS) monitored the displacement continuously during the loading and un-loading measurement. In subsequent clinical investigations, optical markers were placed in the same way on three patients receiving SRS for brain metastasis and enrolled on a prospective clinical trial investigating sources of geometric error in SRS. The OTS monitored the relative position of the patients’ head to the LCF during couch height adjustment within a comfortable range as reported by the patient. The adjustments at the head position (behind the patient neck) were estimated by recording displacements at the foot of the tilting couch. Results: CT-based phantom measurements revealed 2mm displacement relative to the LCF. This error was corrected when the mass was unloaded. OTS-based phantom measures indicated a similar displacement of 1.4mm, and 0.7mm in the posterior and superior directions by loading, and this displacement was corrected by unloading of mass. The range of couch height adjustments for three patients were 4, 12, and 13mm. OTS measurements revealed displacement of the patient’s head relative to the LCF with couch height adjustments in all 3 patients. 3D vector displacement for each patient was a mean (max, in mm) of 0.3 (1.0), 0.5 (1.5), and 0.8 (2.3) during the couch translations. Conclusions: Our results demonstrate that localization of the LCF is accurate and reproducible under constant loading conditions, but uncertainties may arise when stresses are applied by indirect loading in clinical conditions due to couch height adjustments. Further clinical evaluations will elucidate the magnitude of uncertainty and effectiveness of mitigating strategies to reduce sources of geometric uncertainties in SRS delivery. Author Disclosure: Y. Cho, None; G. Bootsma, None; M. Ruschin, None; D. MacFadden, None; M. Hodaie, None; C. Me´nard, None; D. Jaffray, None.
3136
Comparison of Lung Cancer Target Definition Strategies in Stereotactic Body Radiation Therapy
1
J. Wu , C. Betzing2, T. He2, J. Tanyi2, A. Srisuthep2, M. Fuss2 1
Swedish Cancer Institute, Seattle, WA, 2Oregon Health & Science University, Portland, OR
Purpose/Objective(s): Due to the tight margins and rapid dose gradients, SBRT of the lung requires highly accurate tumor volume definition. In this study, we investigated various target definition strategies and their dosimetric impacts. Materials/Methods: Seven patients who received SBRT for lung cancer were studied. For each patient, a free breathing (FB) CT and a 4D CT were acquired. Following the scans, a maximum intensity projection (MIP), an average intensity projection (AIP) and a slow CT (SCT) images were reconstructed. Gross target volumes (GTVs) delineated on the FB, MIP, AIP, and SCT images were compared with the internal target volume (ITV) produced with the union of GTVs delineated on the 4D CT. Three metrics were used for comparison: volume, overlap index (OI), and root mean squared (RMS) distance. To further investigate the dosimetric impact of these contouring strategies, five SBRT plans were created on the FB CT based on the GTVs and ITV delineated above. Planning target volumes (PTVs) were created by adding a 0.5 cm transverse plus a 1.0 cm superior inferior margin to the GTVFB, GTVAIP, and GTVSCT, or a uniform 0.5 cm margin to the ITV and GTVMIP. The prescribed dose was 60 Gy over 3 fractions to the 85% isodose line. For each plan, the corresponding 4D dose was calculated using deformable image registration. The 4D doses were analyzed and compared in terms of tumor D100 (minimum dose received by 100% of GTV) and lung V20 (volume receiving $ 20 Gy). Results: On average the volumes of ITV, GTVMIP, GTVAIP, and GTVSCT were 1.7 ± 0.5 (p = 0.03), 1.3 ± 0.3 (p = 0.43), 0.9 ± 0.2 (p = 0.28), and 0.9 ± 0.2 (p = 0.36) times that of GTVFB. The corresponding OI and RMS were 57%, 57%, 58%, 60% and 0.4, 0.3, 0.4, 0.4 cm, respectively, with respect to GTVFB. The volumes of the corresponding PTVs were 1.1 ± 0.2 (p = 0.89), 0.9 ± 0.2 (p = 0.21), 0.9 ± 0.1 (p = 0.20), and 0.9 ± 0.1 (p = 0.29) times that of PTVFB. For the 4D dose, the tumor D100 of the ITV and GTVMIP based plans was 3.0 ± 4.0 Gy (p = 0.09) and 0.9 ± 4.5 Gy (p = 0.61) above that of the GTVFB based plan, while the tumor D100 of the GTVAIP and GTVSCT based plans was 2.8 ± 6.0 Gy (p = 0.26) and 0.8 ± 4.1 Gy (p = 0.61) below that of the GTVFB based plan. Compared with the GTVFB based plan, the total lung V20 of the ITV based plan was 0.4 ± 1.0% (p = 0.36) absolute higher, while that of the GTVMIP, GTVAIP, and GTVSCT based plans was 0.5 ± 0.7% (p = 0.09), 0.4 ± 0.7% (p = 0.17), and 0.2 ± 0.5% (p = 0.44) absolute lower. Conclusions: Target volumes delineated with the knowledge of tumor motion (ITV, GTVMIP) are larger than those delineated without that information (GTVFB, GTVAIP, GTVSCT). However the margin used to create the PTV for the former cases is smaller than that used for the latter cases. Therefore, even though target definitions depend on imaging protocols, the difference in tumor and lung dose coverage is insignificant. Author Disclosure: J. Wu, None; C. Betzing, None; T. He, None; J. Tanyi, None; A. Srisuthep, None; M. Fuss, None.
3137
Image Guided Stereotactic Radiosurgery (SRS) Treatment of Multiple Brain Metastases using Volumetric Modulated Arc Therapy (VMAT)
K. Smith, Y. Le, E. Ford, T. McNutt, E. Tryggestad, J. Wong Johns Hopkins University, Radiation Oncology and Molecular Radiation Sciences, Baltimore, MD Purpose/Objective(s): With on-board cone-beam CT (CBCT) guidance, invasive head frame is no longer necessary for target localization in the linac-based SRS procedure, but should be retained for rigid immobilization at time of treatment. The procedure allows pre-treatment plan optimization. In this effort, we investigate combining CBCT guidance with VMAT to achieve accurate and efficient SRS of multiple brain metastases. Materials/Methods: Planning studies of a case with two brain metastases were used to compare VMAT with Gamma Knife (GK) and IMRT deliveries. A single, 356 dynamic VMAT plan was generated using the prototype SmartArc feature in the Pinnacle treatment planning system. The corresponding GK and 8-field IMRT plans were also created. All plans prescribed 18 Gy to the PTV. The GK plan was normalized to 50% with 0mm PTV expansion. The VMAT and IMRT plans were normalized to 80%, with 3mm GTV to PTV expansion. A head phantom with an imbedded 9/1600 diameter steel-ball target was used to test delivery accuracy. CT-simulation and treatment planning of the head phantom was performed with Pinnacle. A virtual frame representing
Proceedings of the 51st Annual ASTRO Meeting the invasive frame as a 3D-surface mesh was used to optimize frame location and to minimize beam interference. The real frame was then placed on the phantom according to the plan coordinates. CBCT-guided positioning of the head phantom on the linac was performed with a couch capable of motion with 6 degree of freedom (6DOF). Eight MV localization images were exposed with a 24mm x 24mm open-field at 4 orthogonal and 4 no-coplanar gantry angles. The displacement vector between the centers of the square field and the ball target on each portal image was used to quantify delivery accuracy. Results: Satisfactory coverage of the two PTVs was achieved for all plans with a Dmin of 19.8Gy, 17.1Gy and 19.6Gy to the GTV for VMAT, GK and IMRT, respectively. The conformality index for the VMAT, GK and IMRT plan were 1.28, 1.93 and 1.29, respectively. The VMAT plans were delivered in 6.6mins. It took greater than 1 hour for the GK plans and 30mins for the IMRT plans. The virtual frame tool aided effective frame placement to minimize beam interference. The actual frame did not introduce serious artifacts on the CBCT images. End-to-end tests of the entire procedure show a target can be irradiated to within 1.5 mm of the planned delivery. Conclusions: With VMAT, SRS treatment of multiple brain mets can be performed with a single isocenter in less than 10 mins. Our virtual frame tool allows treatment optimization prior to the day of treatment. Immobilization with the invasive frame ensures highly accurate radiation delivery and also eliminates the need for frequent treatment interruption to monitor or correct intra-fraction motion. Author Disclosure: K. Smith, None; Y. Le, None; E. Ford, None; T. McNutt, Phillips Medical Systems, B. Research Grant; E. Tryggestad, None; J. Wong, Elekta Inc, B. Research Grant; Royalty from Elekta, G. Other.
3138
Improved Accuracy for Non-spine Stereotactic Body Radiation Therapy using Fiducial-based Imageguidance
P. P. Lee1, M. L. Steinberg1, M. T. Selch1, F. G. Abtin2, R. D. Suh2, S. T. Kee2, C. Loh2, N. Agazaryan1 1
UCLA, Department of Radiation Oncology, Los Angeles, CA, 2UCLA, Department of Radiology, Los Angeles, CA
Purpose/Objective(s): Accuracy is essential in stereotactic body radiation therapy (SBRT) to maximize tumor control while minimizing toxicities. A fiducial-based approach offers higher tumor localization accuracy but requires an invasive procedure and delays therapy. We quantified the improvements in patient set-up using gold fiducial markers compared with using spine anatomy in linear accelerator-based SBRT. Materials/Methods: Between August 2008 and March 2009, 16 consecutive patients were treated with 71 fractions (3-5 fractions per tumor, 5-18 Gy per fraction) of fiducial-guided SBRT. Internal tumor volume was defined for all patients based on a 4-Dimensional treatment planning CT. Additional margin (0-6 mm) was used to create a planning tumor volume. There were 30, 21, and 20 fractions treated for tumors of the thorax, abdomen, and pelvis, respectively. Paired oblique kV imaging of fiducial markers was used for initial set-up. Retrospectively, daily set-up was determined independently using either fiducial-to-fiducial matching (DRR to kV imaging) or matching of surrounding stable spine anatomy. Using fiducial-based matching as baseline, set-up error was defined as the absolute displacement for each fraction between fiducial-matching compared to spine-matching in the anterior-posterior (AP), lateral (LAT), and superior-inferior (SI) dimensions. Results: Tumors of the thorax had a mean set-up error in the AP, LAT, and SI dimensions of 4.7 +/- 2.8 mm, 3.2 +/- 3.0 mm, and 4.0 +/- 2.9 mm, respectively. Tumors in the abdomen had a mean set-up error of 5.2 +/- 4.5 mm, 3.0 +/- 2.3 mm, and 5.3 +/- 5.2 mm while tumors in the pelvis had an error of 2.5 +/- 1.6 mm, 2.5 +/- 1.6 mm, and 1.6 +/- 1.0 mm in the same dimensions, respectively. In the AP dimension, 23, 43, and 60% of the fractions in the thorax, abdomen, and pelvis had set-up errors \ 3 mm. In the same sites, this range of error was seen in 60, 57, 75% and 40, 43, 90% in the LAT and SI dimensions, respectively. In the AP dimension, 50, 29, 40% of the fractions in the thorax, abdomen, and pelvis had set-up errors between 3-6 mm. This was seen in 27, 29, 25% and 13, 29, 10% in the LAT and SI dimensions, respectively. Finally, in the AP dimension, 27, 29, and 0% of the fractions in the thorax, abdomen, and pelvis had errors . 6 mm. This was seen in 13, 14, 0% and 13, 29, and 0% in the LAT and SI dimensions, respectively. Conclusions: In the thorax and abdomen, where there is relatively larger motion due to respiration, fiducial markers significantly improve tumor localization during daily set-up. Less set-up errors were detected when tumors in the pelvis were localized using neighboring spine instead of fiducial markers, which may obviate their need in the pelvis. Due to limited sample size, additional data will be needed to confirm these results. Author Disclosure: P.P. Lee, None; M.L. Steinberg, None; M.T. Selch, None; F.G. Abtin, None; R.D. Suh, None; S.T. Kee, None; C. Loh, None; N. Agazaryan, None.
3139
Implementation and Initial Experience with a Non-invasive Image Guided Radiosurgery System for IntraCranial Lesions
F. L. Hacker1,2, S. E. Weiss1,2, K. J. Marcus3,2, N. Ramakrishna4 Dana-Farber Cancer Institute and Brigham & Women’s Hospital, Boston, MA, 2Harvard Medical School, Boston, MA, Dana-Farber Cancer Institute and Brigham & Women’s Hospital and Children’s Hospital Boston, Boston, MA, 4MD Anderson Cancer Center Orlando, Orlando, FL
1
3
Purpose/Objective(s): Intra-cranial stereotactic radiosurgery (SRS) traditionally uses an invasive frame for patient localization and immobilization. We report results of a comprehensive study of localization and immobilization with a system which eliminates the invasive frame by using image guidance for localization and a thermo-plastic mask for immobilization. Based on this study we have implemented the system clinically and we also report on our initial patient experience. Materials/Methods: A stereoscopic kilovoltage x-ray system combined with infra-red (IR) position tracking (ExacTrac by BrainLAB) is used for patient localization. Automatic bony anatomy fusion is used for positioning. Patient immobilization is achieved using a relocatable stereotactic frame with aquaplast mask (BrainLAB Mask System). End to end phantom tests were performed evaluating overall system accuracy. A Rando head phantom with three hidden targets was used. A total of 57 phantom setups were
S693
S692
Volume 75, Number 3, Supplement, 2009
displacement relative to the LCF. The optical tracking system (OTS) monitored the displacement continuously during the loading and un-loading measurement. In subsequent clinical investigations, optical markers were placed in the same way on three patients receiving SRS for brain metastasis and enrolled on a prospective clinical trial investigating sources of geometric error in SRS. The OTS monitored the relative position of the patients’ head to the LCF during couch height adjustment within a comfortable range as reported by the patient. The adjustments at the head position (behind the patient neck) were estimated by recording displacements at the foot of the tilting couch. Results: CT-based phantom measurements revealed 2mm displacement relative to the LCF. This error was corrected when the mass was unloaded. OTS-based phantom measures indicated a similar displacement of 1.4mm, and 0.7mm in the posterior and superior directions by loading, and this displacement was corrected by unloading of mass. The range of couch height adjustments for three patients were 4, 12, and 13mm. OTS measurements revealed displacement of the patient’s head relative to the LCF with couch height adjustments in all 3 patients. 3D vector displacement for each patient was a mean (max, in mm) of 0.3 (1.0), 0.5 (1.5), and 0.8 (2.3) during the couch translations. Conclusions: Our results demonstrate that localization of the LCF is accurate and reproducible under constant loading conditions, but uncertainties may arise when stresses are applied by indirect loading in clinical conditions due to couch height adjustments. Further clinical evaluations will elucidate the magnitude of uncertainty and effectiveness of mitigating strategies to reduce sources of geometric uncertainties in SRS delivery. Author Disclosure: Y. Cho, None; G. Bootsma, None; M. Ruschin, None; D. MacFadden, None; M. Hodaie, None; C. Me´nard, None; D. Jaffray, None.
3136
Comparison of Lung Cancer Target Definition Strategies in Stereotactic Body Radiation Therapy
1
J. Wu , C. Betzing2, T. He2, J. Tanyi2, A. Srisuthep2, M. Fuss2 1
Swedish Cancer Institute, Seattle, WA, 2Oregon Health & Science University, Portland, OR
Purpose/Objective(s): Due to the tight margins and rapid dose gradients, SBRT of the lung requires highly accurate tumor volume definition. In this study, we investigated various target definition strategies and their dosimetric impacts. Materials/Methods: Seven patients who received SBRT for lung cancer were studied. For each patient, a free breathing (FB) CT and a 4D CT were acquired. Following the scans, a maximum intensity projection (MIP), an average intensity projection (AIP) and a slow CT (SCT) images were reconstructed. Gross target volumes (GTVs) delineated on the FB, MIP, AIP, and SCT images were compared with the internal target volume (ITV) produced with the union of GTVs delineated on the 4D CT. Three metrics were used for comparison: volume, overlap index (OI), and root mean squared (RMS) distance. To further investigate the dosimetric impact of these contouring strategies, five SBRT plans were created on the FB CT based on the GTVs and ITV delineated above. Planning target volumes (PTVs) were created by adding a 0.5 cm transverse plus a 1.0 cm superior inferior margin to the GTVFB, GTVAIP, and GTVSCT, or a uniform 0.5 cm margin to the ITV and GTVMIP. The prescribed dose was 60 Gy over 3 fractions to the 85% isodose line. For each plan, the corresponding 4D dose was calculated using deformable image registration. The 4D doses were analyzed and compared in terms of tumor D100 (minimum dose received by 100% of GTV) and lung V20 (volume receiving $ 20 Gy). Results: On average the volumes of ITV, GTVMIP, GTVAIP, and GTVSCT were 1.7 ± 0.5 (p = 0.03), 1.3 ± 0.3 (p = 0.43), 0.9 ± 0.2 (p = 0.28), and 0.9 ± 0.2 (p = 0.36) times that of GTVFB. The corresponding OI and RMS were 57%, 57%, 58%, 60% and 0.4, 0.3, 0.4, 0.4 cm, respectively, with respect to GTVFB. The volumes of the corresponding PTVs were 1.1 ± 0.2 (p = 0.89), 0.9 ± 0.2 (p = 0.21), 0.9 ± 0.1 (p = 0.20), and 0.9 ± 0.1 (p = 0.29) times that of PTVFB. For the 4D dose, the tumor D100 of the ITV and GTVMIP based plans was 3.0 ± 4.0 Gy (p = 0.09) and 0.9 ± 4.5 Gy (p = 0.61) above that of the GTVFB based plan, while the tumor D100 of the GTVAIP and GTVSCT based plans was 2.8 ± 6.0 Gy (p = 0.26) and 0.8 ± 4.1 Gy (p = 0.61) below that of the GTVFB based plan. Compared with the GTVFB based plan, the total lung V20 of the ITV based plan was 0.4 ± 1.0% (p = 0.36) absolute higher, while that of the GTVMIP, GTVAIP, and GTVSCT based plans was 0.5 ± 0.7% (p = 0.09), 0.4 ± 0.7% (p = 0.17), and 0.2 ± 0.5% (p = 0.44) absolute lower. Conclusions: Target volumes delineated with the knowledge of tumor motion (ITV, GTVMIP) are larger than those delineated without that information (GTVFB, GTVAIP, GTVSCT). However the margin used to create the PTV for the former cases is smaller than that used for the latter cases. Therefore, even though target definitions depend on imaging protocols, the difference in tumor and lung dose coverage is insignificant. Author Disclosure: J. Wu, None; C. Betzing, None; T. He, None; J. Tanyi, None; A. Srisuthep, None; M. Fuss, None.
3137
Image Guided Stereotactic Radiosurgery (SRS) Treatment of Multiple Brain Metastases using Volumetric Modulated Arc Therapy (VMAT)
K. Smith, Y. Le, E. Ford, T. McNutt, E. Tryggestad, J. Wong Johns Hopkins University, Radiation Oncology and Molecular Radiation Sciences, Baltimore, MD Purpose/Objective(s): With on-board cone-beam CT (CBCT) guidance, invasive head frame is no longer necessary for target localization in the linac-based SRS procedure, but should be retained for rigid immobilization at time of treatment. The procedure allows pre-treatment plan optimization. In this effort, we investigate combining CBCT guidance with VMAT to achieve accurate and efficient SRS of multiple brain metastases. Materials/Methods: Planning studies of a case with two brain metastases were used to compare VMAT with Gamma Knife (GK) and IMRT deliveries. A single, 356 dynamic VMAT plan was generated using the prototype SmartArc feature in the Pinnacle treatment planning system. The corresponding GK and 8-field IMRT plans were also created. All plans prescribed 18 Gy to the PTV. The GK plan was normalized to 50% with 0mm PTV expansion. The VMAT and IMRT plans were normalized to 80%, with 3mm GTV to PTV expansion. A head phantom with an imbedded 9/1600 diameter steel-ball target was used to test delivery accuracy. CT-simulation and treatment planning of the head phantom was performed with Pinnacle. A virtual frame representing
Proceedings of the 51st Annual ASTRO Meeting the invasive frame as a 3D-surface mesh was used to optimize frame location and to minimize beam interference. The real frame was then placed on the phantom according to the plan coordinates. CBCT-guided positioning of the head phantom on the linac was performed with a couch capable of motion with 6 degree of freedom (6DOF). Eight MV localization images were exposed with a 24mm x 24mm open-field at 4 orthogonal and 4 no-coplanar gantry angles. The displacement vector between the centers of the square field and the ball target on each portal image was used to quantify delivery accuracy. Results: Satisfactory coverage of the two PTVs was achieved for all plans with a Dmin of 19.8Gy, 17.1Gy and 19.6Gy to the GTV for VMAT, GK and IMRT, respectively. The conformality index for the VMAT, GK and IMRT plan were 1.28, 1.93 and 1.29, respectively. The VMAT plans were delivered in 6.6mins. It took greater than 1 hour for the GK plans and 30mins for the IMRT plans. The virtual frame tool aided effective frame placement to minimize beam interference. The actual frame did not introduce serious artifacts on the CBCT images. End-to-end tests of the entire procedure show a target can be irradiated to within 1.5 mm of the planned delivery. Conclusions: With VMAT, SRS treatment of multiple brain mets can be performed with a single isocenter in less than 10 mins. Our virtual frame tool allows treatment optimization prior to the day of treatment. Immobilization with the invasive frame ensures highly accurate radiation delivery and also eliminates the need for frequent treatment interruption to monitor or correct intra-fraction motion. Author Disclosure: K. Smith, None; Y. Le, None; E. Ford, None; T. McNutt, Phillips Medical Systems, B. Research Grant; E. Tryggestad, None; J. Wong, Elekta Inc, B. Research Grant; Royalty from Elekta, G. Other.
3138
Improved Accuracy for Non-spine Stereotactic Body Radiation Therapy using Fiducial-based Imageguidance
P. P. Lee1, M. L. Steinberg1, M. T. Selch1, F. G. Abtin2, R. D. Suh2, S. T. Kee2, C. Loh2, N. Agazaryan1 1
UCLA, Department of Radiation Oncology, Los Angeles, CA, 2UCLA, Department of Radiology, Los Angeles, CA
Purpose/Objective(s): Accuracy is essential in stereotactic body radiation therapy (SBRT) to maximize tumor control while minimizing toxicities. A fiducial-based approach offers higher tumor localization accuracy but requires an invasive procedure and delays therapy. We quantified the improvements in patient set-up using gold fiducial markers compared with using spine anatomy in linear accelerator-based SBRT. Materials/Methods: Between August 2008 and March 2009, 16 consecutive patients were treated with 71 fractions (3-5 fractions per tumor, 5-18 Gy per fraction) of fiducial-guided SBRT. Internal tumor volume was defined for all patients based on a 4-Dimensional treatment planning CT. Additional margin (0-6 mm) was used to create a planning tumor volume. There were 30, 21, and 20 fractions treated for tumors of the thorax, abdomen, and pelvis, respectively. Paired oblique kV imaging of fiducial markers was used for initial set-up. Retrospectively, daily set-up was determined independently using either fiducial-to-fiducial matching (DRR to kV imaging) or matching of surrounding stable spine anatomy. Using fiducial-based matching as baseline, set-up error was defined as the absolute displacement for each fraction between fiducial-matching compared to spine-matching in the anterior-posterior (AP), lateral (LAT), and superior-inferior (SI) dimensions. Results: Tumors of the thorax had a mean set-up error in the AP, LAT, and SI dimensions of 4.7 +/- 2.8 mm, 3.2 +/- 3.0 mm, and 4.0 +/- 2.9 mm, respectively. Tumors in the abdomen had a mean set-up error of 5.2 +/- 4.5 mm, 3.0 +/- 2.3 mm, and 5.3 +/- 5.2 mm while tumors in the pelvis had an error of 2.5 +/- 1.6 mm, 2.5 +/- 1.6 mm, and 1.6 +/- 1.0 mm in the same dimensions, respectively. In the AP dimension, 23, 43, and 60% of the fractions in the thorax, abdomen, and pelvis had set-up errors \ 3 mm. In the same sites, this range of error was seen in 60, 57, 75% and 40, 43, 90% in the LAT and SI dimensions, respectively. In the AP dimension, 50, 29, 40% of the fractions in the thorax, abdomen, and pelvis had set-up errors between 3-6 mm. This was seen in 27, 29, 25% and 13, 29, 10% in the LAT and SI dimensions, respectively. Finally, in the AP dimension, 27, 29, and 0% of the fractions in the thorax, abdomen, and pelvis had errors . 6 mm. This was seen in 13, 14, 0% and 13, 29, and 0% in the LAT and SI dimensions, respectively. Conclusions: In the thorax and abdomen, where there is relatively larger motion due to respiration, fiducial markers significantly improve tumor localization during daily set-up. Less set-up errors were detected when tumors in the pelvis were localized using neighboring spine instead of fiducial markers, which may obviate their need in the pelvis. Due to limited sample size, additional data will be needed to confirm these results. Author Disclosure: P.P. Lee, None; M.L. Steinberg, None; M.T. Selch, None; F.G. Abtin, None; R.D. Suh, None; S.T. Kee, None; C. Loh, None; N. Agazaryan, None.
3139
Implementation and Initial Experience with a Non-invasive Image Guided Radiosurgery System for IntraCranial Lesions
F. L. Hacker1,2, S. E. Weiss1,2, K. J. Marcus3,2, N. Ramakrishna4 Dana-Farber Cancer Institute and Brigham & Women’s Hospital, Boston, MA, 2Harvard Medical School, Boston, MA, Dana-Farber Cancer Institute and Brigham & Women’s Hospital and Children’s Hospital Boston, Boston, MA, 4MD Anderson Cancer Center Orlando, Orlando, FL
1
3
Purpose/Objective(s): Intra-cranial stereotactic radiosurgery (SRS) traditionally uses an invasive frame for patient localization and immobilization. We report results of a comprehensive study of localization and immobilization with a system which eliminates the invasive frame by using image guidance for localization and a thermo-plastic mask for immobilization. Based on this study we have implemented the system clinically and we also report on our initial patient experience. Materials/Methods: A stereoscopic kilovoltage x-ray system combined with infra-red (IR) position tracking (ExacTrac by BrainLAB) is used for patient localization. Automatic bony anatomy fusion is used for positioning. Patient immobilization is achieved using a relocatable stereotactic frame with aquaplast mask (BrainLAB Mask System). End to end phantom tests were performed evaluating overall system accuracy. A Rando head phantom with three hidden targets was used. A total of 57 phantom setups were
S693