Dosimetric Consequences of Deformable Image Registration Errors for Lung SBRT Patients
H Zhong*, J Kim, C Glide-Hurst, I Chetty, Henry Ford Health System, Detroit, MITU-C-141-11 Tuesday 10:30AM - 12:30PM Room: 141
Purpose: To quantify the consequence of deformable image registration (DIR) errors on the total dose calculated for lung SBRT patients.
Methods: 4DCT images were acquired from three lung SBRT patients who had lesions in the upper, middle and lower lobes of lungs. From their end-inhale (EI) CT image, finite element models were developed to generate different sets of displacement vector fields (FEM-DVFs). With these DVFs, mid- and end-exhale (EE) phase images were created from the EI images using a subvoxel-based, mass mapping method. The simulated images were registered to the EI image using Elastix software, and the resultant DVFs were compared with the FEM-DVFs to quantify registration errors at each image voxel. IMRT plans were developed with Pinnacle on simulated maximum intensity projection (MIP) images, and recalculated on each simulated phase image. The resultant dose was warped back to the EI image using both FEM and Elastix DVFs. Relative dose error was quantified via the difference between the two warped doses divided by the prescribed target dose. The dosimetric impact index was then calculated as the ratio of the relative dose error to the registration displacement error at each image voxel.
Results: The mean registration errors evaluated in PTV for the three patients were 1.6, 1.9 and 1.7 mm, and their corresponding dosimetric impact indices were 4.2, 6.1 and 4.9%, respectively. The dosimetric impact indices averaged in PTV, ITV, lung and patient volume for these patients were 5.1, 4.4, 0.4 and 0.7%, and their corresponding standard deviations were 8.1, 7.0, 3.3 and 4.3%, respectively.
Conclusion: The impact of DIR errors on dose accumulation has been quantified with patient-specific FEM-based breathing models. The dosimetric impact indices derived could help assess the quality of dose mapping for lung SBRT patients.
Funding Support, Disclosures, and Conflict of Interest: The research was supported by NIH/NCI R01CA140341.