Abstract:Purpose: To investigate the feasibility of using region-based principal component analysis (PCA) for motion modeling to estimate on-board 4D-CBCT using prior information and limited angle projections for potential 4D target verification of lung radiotherapy. Methods: A technique for lung 4D-CBCT reconstruction has been previously developed using a deformation field map (DFM)-based strategy. In this method, each phase of the 4D-CBCT is generated by deforming a prior CT volume. The DFM is solved by a motion modeling based on global PCA and free-form deformation (MM-FD) technique, using data fidelity constraint and the deformation energy minimization. For this study, a new region-based PCA method was developed to optimize the motion-modeling part of the MM-FD technique in order to estimate target delineation more accurately using fewer on-board projection angles. First, the DFMs are estimated using the region-based PCA divides the motion model extracted from planning 4D-CT into two regions: tumor and everything else. The motion model parameters of both regions are optimized together based on data fidelity constraint with relative weighting added to the tumor PCA model. The 4D digital extended cardiac-torso phantom was used to evaluate the region-based PCA MM-FD technique. A lung patient with a 30 mm diameter lesion was simulated with various anatomical and respirational changes from planning 4D CT to onboard volume. The reconstruction accuracy was evaluated by the Volume-Percent-Difference(VPD)/Center-of-Mass-Shift(COMS) between lesions in the estimated and “ground-truth” on board 4D CBCT. Results: For patient scenarios with various shifts of tumor from planning CT to on-board acquisition with 15 degree orthogonal projections, the average VPD/COMS for the MM-FD technique was 5.04%/0.26 mm and 23.23%/2.4mm for region-based and global PCA MM-FD, respectively. Conclusion: The region-based PCA MM-FD technique can potentially further reduces the scan angle needed for onboard 4D CBCT reconstruction for ultra-fast 4D verification. National Institutes of Health Grant No. R01-CA184173 Varian Medical System

Inter-phase 4D Cardiac MRI Registration with a Motion Prior Derived from CTA

Non-rigid MRI-CT Image Registration with Unsupervised Deep Learning-Based Deformation Prediction

A Comprehensive Cardiac Motion Estimation Framework Using Both Untagged and 3-D Tagged MR Images Based on Nonrigid Registration

Deformable Image Registration Using Vision Transformers for Cardiac Motion Estimation from Cine Cardiac MRI Images

Deep-learning-based groupwise registration for motion correction of cardiac $T_1$ mapping

Robust Fast Inter-Bin Image Registration for Undersampled Coronary MRI Based on a Learned Motion Prior

MDH-Net: Advancing 3D Brain MRI Registration with Multi-Stage Transformer and Dual-Stream Feature Refinement Hybrid Network

4D Cone-beam CT Deformable Registration using Unsupervised Spatial Transformation Network

Respiratory Deformation Registration in 4D-Ct/cone Beam CT Using Deep Learning.

Motion Correction for Native Myocardial T1 Mapping Using Self-Supervised Deep Learning Registration with Contrast Separation.

4d Ultrasound And 3d Mri Registration Of Beating Heart

Artifacts Reduction in 4D-CBCT Via a Joint Free-form Registration Method of Projection Match and Gradient Constraint

Motion-Guided Deep Image Prior for Cardiac MRI

TH-CD-303-06: Deformable Registration-Based Image Estimation Method for 4D CBCT Using Region-Based PCA

Motion Decoupling and Registration for 3D Magnetic Resonance Myocardial Perfusion Imaging

Biomechanics-informed Neural Networks for Myocardial Motion Tracking in MRI

Experimental Research of in Vivo Mouse Cardiac 4D Micro-CT Imaging Via Deformation Vector Field Registration

Sketch-based Registration of 3D Cine MRI to 4D flow MRI

4D-Ct Motion Estimation Using Deformable Image Registration and 5d Respiratory Motion Modeling

Cardiac Motion and Deformation Estimation from Tagged MRI Sequences Using a Temporal Coherent Image Registration Framework

3D/2D model-to-image registration by imitation learning for cardiac procedures