Room: Karl Dean Ballroom C
Purpose: To develop a new motion modeling technique to account for the deformation caused by the head, neck, and shoulder posture changes.
Methods: A patient-specific physical motion model was reconstructed with the following steps: (a) segmenting the bony voxels from CT using Hounsfield unit (HU) threshold; (b) clustering the bony voxels and assigning the clusters to individual bones by pattern matching; (c) dividing the spine into vertebrae by searching for the intervertebral discs; (d) rigidly registering each bone (and each vertebra) from two CTs using iterative closest point algorithm; (e) segmenting soft tissue using HU threshold and clustering; (f) deforming soft tissue with bones and refining the deformation using finite element analysis. Eighteen patients from the Cancer Imaging Archive were selected to validate the technique. Each patient had two PETCTs and a planning CT taken with different postures. Motion modeling was performed to deform the PETCTs to the planning CT. The deformation maps derived from motion modeling were used for quality assurance (QA) of a commercial deformable image registration (DIR) software.
Results: The physical modeling technique was able to register the individual bones and deform the soft tissue for all patients, including those with large posture changes. For the commercial DIR software, mean errors for the head, jaw, spine, and sternum were 0.60, 0.28, 0.53, 0.30, and 0.39 cm respectively. Mean errors for the clavicles, scapulae and humerus were 0.36, 0.54 and 1.38 cm for patients with arms down in all scans and 3.17, 3.09 and 6.83 cm for patients with arms up in PETCTs and down in the planning CT.
Conclusion: We developed a patient-specific motion modeling technique that is capable of measuring the deformation caused by head, neck and shoulder position changes. The modeling technique can be used as a QA tool of commercial DIR software.
Funding Support, Disclosures, and Conflict of Interest: Ping Xia receives funding support from Philips, Inc.