Author: Chih-Chiang Chang, Chingyun Cheng, Ben Durkee, Minglei Kang, Elissa Khoudary, Yangguang Ma, Xuanqin Mou, YuFei Wang, Tengda Zhang, Wang Zhengda π¨βπ¬
Affiliation: Department of Human Oncology, University of Wisconsin School of Medicine and Public Health, Department of Radiation Oncology, The First Affiliated Hospital of Zhengzhou University, School of Software Engineering, Xiβan Jiaotong University, School of information and communications engineering, Faculty of electronic and information engineering, Xiβan Jiaotong University, Department of Medical Physics, Columbia University, University of Pennsylvania, Department of Human Oncology, University of Wisconsin-Madison π
Purpose: This study presents a voxel-based analysis method based on deformable image registration to accurately quantify respiratory-induced motion and deformation of both targets and organs at risk (OARs) using 4DCT. This approach provided critical insights to guide decision-making and optimize motion mitigation strategies.
Methods: A voxel-based statistical analysis was conducted using deformable registration implemented through an enhanced non-parametric algorithm within an open-source tool. Organ motion and volumetric variations across individual phases were quantified using voxel-based displacement and water-equivalent thickness variations (ΞWET), with the inhale phase serving as the reference. A motion index (MI) was defined as the ratio of voxels with ΞWET greater than 5 mm to the total number of voxels within the region of interest (ROI). T-tests and Pearson correlations were performed, with a significance threshold set at p < 0.05. 4DCT datasets of ten respiratory phases from 11 lung cancer patients were analyzed and the cohort was stratified by internal target volume (ITV) location into left upper (n=4), left lower (n=3), and right lower lobe (n=4) subgroups.
Results: The superior-inferior displacement was 2-6 times greater than other motion directions. The center and voxel-averaged displacements exhibit no significant differences observed across individual ROIs. Volumetric deformation was highest in the diaphragm and mediastinum (60-80%), with ITV deformation varying from <5% to 70%. Maximum ΞWET and MI sensitivity occurred at 50% (7 cases), 60% (3 cases), and 70% (1 case) respiratory phases. ITV and OARs motion parameters (center motion, voxel-averaged motion, 80th/90th-percentile motion) showed strong correlations with each other (r = 0.85β1.00, p< 0.05). The ROIs, including various lung lobes, the diaphragm, and the mediastinum, showed strong correlations (p-value<0.01) between voxel-averaged motion patterns and ITV motion.
Conclusion: This method provides a comprehensive quantification of respiratory motion and deformation, offering valuable insights for optimizing motion management strategies in lung cancer radiotherapy.