Main Session
Sep
28
PQA 01 - Radiation and Cancer Physics, Sarcoma and Cutaneous Tumors
2211 - Evaluation of the Targeting and Dosimetric Accuracy of Biomechanical Model-Based Single Isocenter Adaptive Radiotherapy Planning for Oligometastatic Bone Metastases
Presenter(s)
Kamal Singhrao, PhD - Brigham and Women's Hospital/Dana Farber Cancer Institute, Boston, MA
R. E. Rodríguez-Pérez1, G. U. Perez Rojas1, B. A. Maldonado Luna1, B. D. C. Alonso1, K. Moerman2, J. S. Bredfeldt3, M. A. Huynh3, and K. Singhrao3; 1Faculty of Mathematical Physical Sciences, Benemerita Universidad Autonoma de Puebla, Puebla, Mexico, 2Department of Mechanical Engineering, University of Galway, Galway, Ireland, 3Department of Radiation Oncology, Brigham and Women's Hospital, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA
Purpose/Objective(s):
Stereotactic body radiotherapy (SBRT) for oligometastatic bone metastases involves treatment of each metastasis site successively (multi-isocenter planning); resulting in long treatment times, increased plan complexity, and potential exclusion of RT as a therapy for patients with numerous metastases.
Treatment of all metastasis sites at once with a single treatment plan (single-isocenter planning) combined with online adaptive RT could address these issues, but this method requires precise target contour delineation. Our standard of care, for online adaptive SBRT, is to rigidly align planned targets on treatment day to mitigate unphysical deformations from intensity-based deformable image registration (DIR).
We propose using a modality-independent finite element analysis-based biomechanical model to provide physically realistic deformed target contours. Target deformation accuracy is assessed and dosimetric accuracy is quantified with a simulated adaptive single isocenter plan.
Materials/Methods:
A male pelvis dataset with 16 patients was used, each with two CT scans (reference and treatment day CTs). For all patients, three tumor targets were drawn on the right femur, left pubis and sacrum. Single and multiple isocenter treatment plans, and corresponding dose maps, were simulated using the MatRad toolbox. The prescribed dose was 30 Gy delivered in 5 fractions.
Contours were generated for targets and organs-at-risk (OAR). To estimate the initial transformations for each contour, rigid registration was applied iteratively until the maximum overlap, computed by the Dice coefficient, was reached. Deformation vector fields (DVFs) were generated using a finite element simulation based on open-source software, FEBio and GIBBON. The DVFs were combined and applied to the treatment day CT.
The Dice coefficient was used to assess target registration accuracy. Simulated adapted single- and multi-isocenter plans were evaluated via the ?-index (3% dose difference, 3mm distance-to-agreement), global mean dose difference (GMD) and local mean dose difference (LMD).
Results:
The mean and maximum centroid displacements across all targets were 0.22±0.06cm and 0.47cm. The Dice statistic for target registration was 0.87±0.03. For single isocenter plans, the acquired GMD were 1.1x10-3±2.4x10-4 and LMD values were 3.1x10-3±7.1x10-3. The ?-index for the single isocenter and multi-isocenter plans were 0.93±0.02 and 0.95±0.01.
Conclusion:
These results show that single isocenter BM-derived target contour produce comparable targeting and dosimetric accuracy to multi-isocenter plans for oligometastatic bone metastases. Future work will involve prospective validation of this method for online adaptive X-ray and MRI-guided SBRT.
Abstract 2211 - Table 1
| Type of plan | ?-index | GMD (Gy) | LMD (Gy) | Dice Coefficient |
| Single Isocenter | 0.93±0.02 | 1.1E-3 ± 2.4E-4 | 3.1.4E-3 ± 7.1E-3 | 0.87 ± 0.03 |
| Multi-isocenter | 0.95±0.01 | 3.9E-3 ± 2.8E-3 | 3.6E-2 ± 2.2e-2 | 0.87 ± 0.03 |