Author: Hua-Chieh Shao, Chenyang Shen, Jiacheng Xie, Shunyu Yan, You Zhang 👨🔬
Affiliation: Medical Artificial Intelligence and Automation (MAIA) Lab, Department of Radiation Oncology, UT Southwestern Medical Center, UT Southwestern Medical Center, Medical Artificial Intelligence and Automation (MAIA) Laboratory, Department of Radiation Oncology, UT Southwestern Medical Center, Medical Artificial Intelligence and Automation (MAIA) Laboratory, Department of Radiation Oncology, UT Southwestern Medical Center, The University of Texas at Dallas 🌍
Purpose: Motion-resolved CBCT imaging, which reconstructs a dynamic sequence of CBCTs reflecting intra-scan motion (one CBCT per x-ray projection), is highly desired for regular/irregular motion characterization, patient setup, and motion-adapted radiotherapy. Representing patient anatomy and associated motion fields as 3D Gaussians, we developed a Gaussian representation-based framework (PMF-STGR) for fast and accurate dynamic CBCT reconstruction.
Methods: PMF-STGR comprises three major components: a dense set of 3D Gaussians to reconstruct a reference-frame CBCT for the dynamic sequence; another 3D Gaussian set to capture three-level, coarse-to-fine motion-basis-components(MBCs) to model the intra-scan motion; and a CNN-based motion encoder to solve projection-specific temporal coefficients for the MBCs. Scaled by the temporal coefficients, the learned MBCs will combine into deformation-vector-fields(DVFs) to deform the reference CBCT into projection-specific CBCTs to capture the dynamic motion. Due to the strong representation power of 3D Gaussians, PMF-STGR can reconstruct dynamic CBCTs in a ‘one-shot’ training fashion from a standard 3D CBCT scan, without using any prior anatomical/motion model. We evaluated PMF-STGR using XCAT phantom simulations and real patient scans. The XCAT study simulates lung CBCT scans for seven free-breathing scenarios featuring various motion irregularities. For patients, we used CBCT projection sets from five cases. Metrics including image relative-error(RE), structural-similarity-index-measure(SSIM), tumor center-of-mass-error(COME), and landmark localization-error(LE) were used to evaluate the accuracy of solved dynamic CBCTs and motion.
Results: PMF-STGR achieved excellent reconstruction accuracy and outperformed a benchmark state-of-the-art technique, PMF-STINR. For XCAT, the mean(±s.d.) RE, SSIM, and COME were 0.123(0.012), 0.991(0.002), and 0.726mm(0.254mm) for PMF-STGR, compared with 0.149(0.016), 0.944(0.006), and 0.935mm(0.178mm) for PMF-STINR. For patients, the mean(±s.d.) landmark LE were 1.235mm(1.132mm) for PMF-STGR, and 1.277mm(1.309mm) for PMF-STINR. Compared with PMF-STINR, PMF-STGR reduced reconstruction time by 40%.
Conclusion: PMF-STGS has demonstrated excellent accuracy/efficiency in reconstructing motion-resolved CBCTs from a standard 3D CBCT scan, paving the way for informed, effective motion management.