Image-based estimation and nonparametric modeling: Towards enhanced geometric calibration of an X-ray system

Geometric calibrations of medical imaging systems are crucial to allow for advanced (X-ray) imaging techniques. Developments in medical procedures, lightweight system design and the growing costs of healthcare, leads to the desire for simpler and faster calibration approaches. The aim of this paper is to present a novel approach to enhance system calibrations for a wide range of imaging applications. The method is based on the introduction of small markers within the line of sight of the system, by virtue of a small mechanical adjustment to the system. By detecting markers in the X-ray images, displacements between the systems X-ray source and detector are in-situ measured. Additionally, the approach can be used to obtain nonparametric models of the dynamics of the mechanical system, enabling advanced observer-based estimation approaches. The potential of the method is illustrated by experimental results.