Breathing and heart beat cause high-amplitude cyclic dislocation and deformation of organs in the human body. Both types of motion are interesting in itself, since they express physiological important functions but they are also the cause of image reconstruction artefacts, location insecurity in interventions and other problems in medical diagnosis and therapy. On the other hand, state of the art medical imaging equipment is able to capture cardiac and breathing motion (e.g. cardiac or breathing gated CT) and allows therefore capturing the patient individual motion patterns. Once a patient individual motion field (i.e. voxel dislocation vectors from one motion phase point to the next one) is extracted, it can be used for motion compensated reconstruction (of e.g. PET, SPECT, and CT images), the extraction of patient individual safety margins for radio therapy planning, or motion compensated minimal invasive interventions. The project is done in cooperation with Philips Research Europe - Hamburg.
A first goal is to develop methods for the extraction of breathing motion fields from medical image data followed by a detailed analysis of the obtained motion fields. The analysis builds the basis for further activities, e.g., developing compact motion representations and motion modeling. From a modeling perspective, one challenging task is to not only focus on patient-specific motion models but also generate inter-subject models.
Motion fields are extracted using a shape constrained deformable model approach being also to account for discontinuities in organ motion but on the same time fast and accurate. Motion model generation is carried out by the use of adequate reference structures.