Robust real-time 3D respiratory motion detection using time-of-flight cameras

Abstract

PurposeRespiratory motion of the patient during data acquisition causes artifacts in the field of emission or computed tomography. Respiratory gating allows to track and correct these artifacts.Materials and methodsIn this paper, we present a system that uses the fairly new and off-the-shelf time-of-flight (ToF) technology to compute a dense estimate of the three-dimensional respiratory motion of a patient. The work is characterized by three key contributions. The first is the employment of ToF sensors. Using ToF sensors, it is feasible to acquire a dense 3D surface model of the chest and abdomen of the patient with more than 15 frames per second. The second contribution is an algorithm to derive a surface representation which enables the estimation of the 3D respiratory motion of the patient, which is sufficient to compute 1D breathing signals for scalable specific regions of interest like chest and abdomen. The proposed data-driven algorithm models the chest and abdomen three-dimensionally by fitting distinct planes to different regions of the torso of the patient. The third contribution is the possibility to derive a sub-millimeter accurate 1D respiratory motion signal by observing the displacement of each plane.ResultsOur ToF modeling approach enables marker less, real-time, 3D tracking of patient respiratory motion with an accuracy of 0.1 mm.ConclusionThus, our approach provides 1D breathing signals for scalable anatomical regions of interest with sufficient accuracy for artifact reduction in SPECT or X-ray angiography.