Photometric Estimation of 3D Surface Motion Fields for Respiration Management

Abstract

In radiation therapy, the estimation of torso deformations due to respiratory motion is an essential component for real-time tumor tracking solutions. Using range imaging (RI) sensors for continuous monitoring during the treatment, the 3-D surface motion field is reconstructed by a non-rigid registration of the patient’s instantaneous body surface to a reference. Typically, surface registration approaches rely on the pure topology of the target. However, for RI modalities that additionally capture photometric data, we expect the registration to benefit from incorporating this secondary source of information. Hence, in this paper, we propose a method for the estimation of 3-D surface motion fields using an optical flow framework in the 2-D photometric domain. In experiments on real data from healthy volunteers, our photometric method outperformed a geometry-driven surface registration by 6.5% and 22.5% for normal and deep thoracic breathing, respectively. Both the qualitative and quantitative results indicate that the incorporation of photometric information provides a more realistic deformation estimation regarding the human respiratory system.