- Email: [email protected]
A Practical Method of Adaptive Radiotherapy for Prostate Cancer using Real-time Electromagnetic Tracking
Proceedings of the 51st Annual ASTRO Meeting Results: For the 27 spine SRT treatments, the (mean ± SD) of the translation residuals was largest along the SI direction: (Lat, SI, AP) = (0.34 ± 0.31, 0.45 ± 0.5, 0.34 ± 0.27) mm. The rotation residuals are the largest around the AP direction: (Lat, SI, AP) = (0.3 ± 0.4, 0.5 ± 0.4, 0.5 ± 0.5)o. Post-Tx total positional errors were (0.65 ± 0.62, 0.78 ± 0.68, 0.34 ± 0.27) mm and (0.5 ± 0.5, 0.8 ± 0.8, 0.7 ± 0.7)o. Intrafraction motion is evaluated by the absolute difference between post-Tx total errors and pre-Tx residual errors : (0.64 ± 0.61, 0.63 ± 0.57, 0.70 ± 0.80) mm, (0.4 ± 0.4, 0.7 ± 0.7, 0.5 ± 0.4). Conclusions: In this evaluation of spine SRT, the intrafraction positional error has been shown to be consistently in the sub-millimeter range, proving that a 2 mm setup planning margin is conservatively appropriate for a frameless spine SRT technique provided 3D image guidance is utilized. Author Disclosure: D. Ionascu, None; S. Martin, None; J. Wloch, None; M. Matuszak, None; I.S. Grills, None; D. Yan, None.
2873
A Practical Method of Adaptive Radiotherapy for Prostate Cancer using Real-time Electromagnetic Tracking
J. R. Olsen, C. Noel, K. Baker, L. Santanam, J. Michalski, P. J. Parikh Washington University School of Medicine, St. Louis, MO Purpose/Objective(s): Current methods of adaptive radiation therapy rely on serial imaging, which creates a segmentation and image analysis workload for the physician. We have created an automated process utilizing real-time electromagnetic tracking data to evaluate the adequacy of PTV margins in prostate cancer, allowing for a ‘physician-less’ process of adaptive radiation therapy. We present an analysis of PTV adequacy, and a validation of our adaptive process. Materials/Methods: Tracking data was analyzed for 15 patients who underwent SMLC IMRT with uniform 5mm margins for prostate cancer utilizing the Calypso 4D Localization System between 11/2007 and 3/2009. Treatment was delivered only if reported setup rotation was less than 10 and while isocenter displacement was less than 3 mm during real-time tracking, to prevent radiation delivery during large anatomical variations. Additional plans were generated with 0 and 3 mm PTV margins. A custom software application that convolved the original dose distribution and structure location from simulation with the over 50,000 measured translations and rotations for each patient’s course of therapy was developed in Matlab. The dose delivered to the prostate was then calculated for the first 10 fractions, and for the entire treatment. A treatment course was considered adequate if the minimum delivered dose to the prostate (Dmin) was at least 98% of the planned Dmin. Results: For 0, 3, and 5 mm PTV margins, adequate treatment was obtained in 3/15, 12/15, and 15/15 patients, and the delivered Dmin ranged from 78-99, 96-100, and 99-100 percent of the planned Dmin. While the degree of translational and rotational motion varied widely among patients, adequacy of treatment during the first 10 fractions predicted sufficient dose delivery for the entire treatment for all patients and margins. For the 15 cases of inadequate treatment delivery, early analysis after the first 10 fractions revealed the need for PTV expansion thereby validating the adaptive process. Conclusions: Our adaptive process successfully utilized real time electromagnetic tracking data to predict the need for PTV modifications, without the added burden of physician contouring and image analysis. While applied here to standard fractionation prostate cancer treatment, our dose verification methods are readily applicable to hypofractionated treatment, and other instances in which real time tracking is utilized. Author Disclosure: J.R. Olsen, None; C. Noel, None; K. Baker, None; L. Santanam, None; J. Michalski, None; P.J. Parikh, Calypso Medical Systems, B. Research Grant.
2874
The Accuracy of Deformable Registration for Adaptive Radiotherapy of Head and Neck Cancer
S. van Kranen, A. Mencarelli, S. van Beek, C. Rasch, J. Sonke, M. van Herk The Netherlands Cancer Institute, Amsterdam, Netherlands Purpose/Objective(s): Currently there is a large interest to use deformable image registration methods for adaptive radiotherapy and dose accumulation. When such tools are used to support clinical decisions, correct validation of the accuracy is essential. We recently implemented two different deformable image registration strategies for head and neck cancer patients. The first is a simple method based on local rigid registration of multiple regions surrounding bony anatomy, interpolated using thin-plate splines. This method is close to clinical implementation. The second method is based on b-splines, and it is more used for research tasks. Both methods employ correlation ratio as driving cost function. The aim of this paper is to rigorously validate both methods. Materials/Methods: Data of 5 patients were analyzed so far. For each patient, the planning CT (2 mm slices) and 2 CBCT scans (Elekta Synergy, 1mm voxel size) have been analyzed. Two validation methods are applied. First, two observers and the deformable registration algorithms identify a set of corresponding anatomical landmarks (30 per scan pair). The uncertainties in both the human and computer observer are derived using analysis of variance, taking the inaccuracies of the human observers into account. Second, small fragments of Visicoil markers (on average 6 per patient) have been implanted on the border of the tumor. The distance of these markers (which are ignored by the registration algorithm) after registration was evaluated. Results: Both deformable registration methods result in visually acceptable registrations. The accuracy of the multi-region method analyzed from the anatomical landmarks was on the order of 1.4 mm SD for all three axes. B-splines performed better on the landmarks: 1.1 mm SD. The accuracy derived from the implanted markers was slightly worse, 1.5 mm SD on average for B-splines. We observed that wound healing processes around the tumor could result in large marker displacements parallel with anatomical borders. The landmark data may be considered representative for tissues with visible borders, i.e., most organs at risk. The implanted markers are representative for tumor borders. Conclusions: Some head and neck tumor borders show considerable migration during external beam therapy. The tested deformable image registration methods have an excellent performance for normal tissues, but are sometimes not capable of capturing the motion of the tumor borders. Caution should therefore be exercised when using such algorithms to evaluate accumulated tumor doses. Author Disclosure: S. van Kranen, None; A. Mencarelli, None; S. van Beek, None; C. Rasch, None; J. Sonke, Elekta Oncology Systems, B. Research Grant; Elekta Oncology Systems, F. Consultant/Advisory Board; M. van Herk, Elekta Oncology Systems, B. Research Grant; Elekta Oncology Systems, F. Consultant/Advisory Board.
S573
2873
A Practical Method of Adaptive Radiotherapy for Prostate Cancer using Real-time Electromagnetic Tracking
J. R. Olsen, C. Noel, K. Baker, L. Santanam, J. Michalski, P. J. Parikh Washington University School of Medicine, St. Louis, MO Purpose/Objective(s): Current methods of adaptive radiation therapy rely on serial imaging, which creates a segmentation and image analysis workload for the physician. We have created an automated process utilizing real-time electromagnetic tracking data to evaluate the adequacy of PTV margins in prostate cancer, allowing for a ‘physician-less’ process of adaptive radiation therapy. We present an analysis of PTV adequacy, and a validation of our adaptive process. Materials/Methods: Tracking data was analyzed for 15 patients who underwent SMLC IMRT with uniform 5mm margins for prostate cancer utilizing the Calypso 4D Localization System between 11/2007 and 3/2009. Treatment was delivered only if reported setup rotation was less than 10 and while isocenter displacement was less than 3 mm during real-time tracking, to prevent radiation delivery during large anatomical variations. Additional plans were generated with 0 and 3 mm PTV margins. A custom software application that convolved the original dose distribution and structure location from simulation with the over 50,000 measured translations and rotations for each patient’s course of therapy was developed in Matlab. The dose delivered to the prostate was then calculated for the first 10 fractions, and for the entire treatment. A treatment course was considered adequate if the minimum delivered dose to the prostate (Dmin) was at least 98% of the planned Dmin. Results: For 0, 3, and 5 mm PTV margins, adequate treatment was obtained in 3/15, 12/15, and 15/15 patients, and the delivered Dmin ranged from 78-99, 96-100, and 99-100 percent of the planned Dmin. While the degree of translational and rotational motion varied widely among patients, adequacy of treatment during the first 10 fractions predicted sufficient dose delivery for the entire treatment for all patients and margins. For the 15 cases of inadequate treatment delivery, early analysis after the first 10 fractions revealed the need for PTV expansion thereby validating the adaptive process. Conclusions: Our adaptive process successfully utilized real time electromagnetic tracking data to predict the need for PTV modifications, without the added burden of physician contouring and image analysis. While applied here to standard fractionation prostate cancer treatment, our dose verification methods are readily applicable to hypofractionated treatment, and other instances in which real time tracking is utilized. Author Disclosure: J.R. Olsen, None; C. Noel, None; K. Baker, None; L. Santanam, None; J. Michalski, None; P.J. Parikh, Calypso Medical Systems, B. Research Grant.
2874
The Accuracy of Deformable Registration for Adaptive Radiotherapy of Head and Neck Cancer
S. van Kranen, A. Mencarelli, S. van Beek, C. Rasch, J. Sonke, M. van Herk The Netherlands Cancer Institute, Amsterdam, Netherlands Purpose/Objective(s): Currently there is a large interest to use deformable image registration methods for adaptive radiotherapy and dose accumulation. When such tools are used to support clinical decisions, correct validation of the accuracy is essential. We recently implemented two different deformable image registration strategies for head and neck cancer patients. The first is a simple method based on local rigid registration of multiple regions surrounding bony anatomy, interpolated using thin-plate splines. This method is close to clinical implementation. The second method is based on b-splines, and it is more used for research tasks. Both methods employ correlation ratio as driving cost function. The aim of this paper is to rigorously validate both methods. Materials/Methods: Data of 5 patients were analyzed so far. For each patient, the planning CT (2 mm slices) and 2 CBCT scans (Elekta Synergy, 1mm voxel size) have been analyzed. Two validation methods are applied. First, two observers and the deformable registration algorithms identify a set of corresponding anatomical landmarks (30 per scan pair). The uncertainties in both the human and computer observer are derived using analysis of variance, taking the inaccuracies of the human observers into account. Second, small fragments of Visicoil markers (on average 6 per patient) have been implanted on the border of the tumor. The distance of these markers (which are ignored by the registration algorithm) after registration was evaluated. Results: Both deformable registration methods result in visually acceptable registrations. The accuracy of the multi-region method analyzed from the anatomical landmarks was on the order of 1.4 mm SD for all three axes. B-splines performed better on the landmarks: 1.1 mm SD. The accuracy derived from the implanted markers was slightly worse, 1.5 mm SD on average for B-splines. We observed that wound healing processes around the tumor could result in large marker displacements parallel with anatomical borders. The landmark data may be considered representative for tissues with visible borders, i.e., most organs at risk. The implanted markers are representative for tumor borders. Conclusions: Some head and neck tumor borders show considerable migration during external beam therapy. The tested deformable image registration methods have an excellent performance for normal tissues, but are sometimes not capable of capturing the motion of the tumor borders. Caution should therefore be exercised when using such algorithms to evaluate accumulated tumor doses. Author Disclosure: S. van Kranen, None; A. Mencarelli, None; S. van Beek, None; C. Rasch, None; J. Sonke, Elekta Oncology Systems, B. Research Grant; Elekta Oncology Systems, F. Consultant/Advisory Board; M. van Herk, Elekta Oncology Systems, B. Research Grant; Elekta Oncology Systems, F. Consultant/Advisory Board.
S573