Program Information
Biomechanical Modeling Assisted Simultaneous Motion Estimation and Image Reconstruction Incorporating for 4D-CBCT
X huang*, Y Zhang , J Wang , UT Southwestern Medical Center, Dallas, TX
Presentations
TH-AB-601-12 (Thursday, August 3, 2017) 7:30 AM - 9:30 AM Room: 601
Purpose: Simultaneous motion estimation and image reconstruction (SMEIR) has shown advantages in reconstructing high-quality four dimensional (4D) CBCT by utilizing an inter-phase motion model that is estimated from the projection data directly. The goal of this work is to improve the accuracy of deformation vector fields (DVFs) estimation at low-contrast regions by incorporating biomechanical modeling of lung.
Methods: The original SMEIR algorithm consists of two alternating steps: 1) DVFs estimation by matching the forward projection of deformed volume and projection measurements using a 2D-3D registration technique; and 2) motion-compensated image reconstruction using projections from all phases to update a single phase 4D-CBCT based on the estimated DVFs. In the proposed biomechanical modeling assisted (SMEIR-Bio) method, the Mooney-Rivlin material model is used to simulate lung motion, where DVFs at lung surface obtained from motion estimation step in SMEIR serve as the boundary condition. The combination of DVFs inside lung from biomechanical modeling and outside lung from SMEIR is used in motion-compensated image reconstruction. The proposed method is evaluated with 4D-CBCT projections simulated from 4D-CT images of 11 lung cancer patients. For each patient, 10, 20, 30 and 40 projections per phase are simulated to mimic different acquisition times.
Results: Using 30 projections per phase, the Mean±SD of normalized cross correlations are 0.953±0.0117 and 0.939±0.0165 for SMEIR-Bio and SMEIR, respectively, compared to the original 4D-CT at selected low-contrast lung region. The average difference of all patients of landmarks displacement between 4D-CT and that extracted from SMEIR-Bio and SMEIR along superior-inferior are 2.49 mm and 4.03 mm, respectively.
Conclusion: By incorporating mechanical modeling of lung, SMEIR-Bio improves DVF estimation accuracy for low-contrast regions of lung, leading to higher quality 4D-CBCT as compared to the original SMEIR algorithm.
Funding Support, Disclosures, and Conflict of Interest: the American Cancer Society (RSG-13-326-01-CCE), the US National Institutes of Health (R01 EB020366), the Cancer Prevention and Research Institute of Texas (RP130109) and Elekta Inc
Contact Email: