TH‐C‐BRC‐11: 3D Tracking of Interfraction and Intrafraction Head and Neck Anatomy during Radiotherapy Using Multiple Kinect Sensors

Abstract

Purpose: The process of daily measurement and validation of tumors and normal structures with onboard imaging provides information useful for reducing patient setup uncertainty errors. However, the use of daily onboard CT imaging greatly increases the radiation dose to critical structures that lie within the CT volume. We now present a quantitative skin surface 3D imaging that when coupled with quantitative patient‐specific biomechanical models determine the tumor and normal organ deformation caused by routine patient head and neck misalignments. Incorporating such modeling and imaging could substantially decrease the number of cone‐ beam CT scans required for patient setup and ultimately the daily CT scan dose. Methods:The quantitative skin surface 3D imaging that monitors the patient anatomy are developed using multiple Kinect sensors. A set of 4 3D cameras are used for illustration purposes to track the patient anatomy externally. Of the 4 cameras, 3 of them are used to track the patientˈs anatomy contour (e.g. face, hands etc) using depth and intensity based contour tracking. Such an approach provides a set of 3D contours for each anatomical region from each camera. The 4th camera employs a marker less face recognition and tracking for delineating the region of the patientˈs face. The location of the face is then shared among the camera controllers in realtime and the anatomical contour that closely matches the face region is selected. Once selected, all the contours are then integrated to form a single 3D representation of the anatomy and overlapping contours are cleaned using Voronoi data re‐sampling. Results: The unified 3D representation presents a quantified 3D surface image within a precision range of 0.3 cm at an acquisition rate of 30 frame‐per‐second. Conclusions :Daily measurement of 3D skin surface with the proposed imaging system is feasible for reducing patient setup uncertainty errors while minimizing radiation exposure risk.