Automated detection of (unintended) patient motion in dental Cone Beam CT (CBCT)

Similar to conventional Computed Tomography (CT) dental Cone Beam Computed Tomography (CBCT) generates 3D-data by means of (filtered) backprojection from a multitude of two-dimensional (2D) projection radiographs acquired in a circular arrangement around the patient’s head. Due to the construction of the devices, this image acquisition takes several (commonly > 10s) seconds. As a consequence, pa-tients frequently move slightly during this period resulting in a decreased image quality. One typical effect is motion blur that reduces the effective spatial resolu-tion of the images. This project aims to automatically detect such motion in the 2D-projection images. CBCT-data produced for pre-operative planning of guided implant insertion require extra high spatial resolution. Commonly the radiographic splints applied for such planning integrate (spherical) radio-opaque markers for re-gistration of the 3D-data to the patient coordinates. Our project starts using the spherical marker projections on the 2D projection radiographs. In addition, general marker-free approaches for motion detection and quantification will be developed.