Improvement and Evaluation of a Time-of-Flight-based Patient Positioning System

In this paper we improve a surface-based patient positioning method. The method describes a system for automatic positioning of patients using time-of-flight cameras in radiotherapy. To improve the registration result three new preprocessing steps (bilateral filtering, temporal averaging, variance filtering) are introduced to the processing pipeline. Furthermore, the accuracy of the surface matching algorithm (ICP) is improved by changing the distance measurement from point-to-point to point-to-surface. The mean registration error is improved by 2.14mm to 0.74mm, whereby the working distance could also be increased from 0.8m to 1.5m.