Segmentation-Enhanced Registration of Angiography Data

For diagnostic and therapeutic purposes, multiple imaging techniques can be used to gain more information on a patient’s anatomy. In maxillofacial and neurosurgery, three-dimensional rotational angiography (3DRA) images can provide a detailed view on blood vessels, whereas computed tomography angiography (CTA) images are common for multiplanar interpretation of bone, soft tissue, and blood vessels. Thus, the registration of 3DRA to CTA allows for a better understanding of the structure and location of blood vessels within a context of bone and tissue, which is essential to surgical interventions. Among other reasons, the lack of mutual information between 3DRA and CTA images makes their registration challenging. In this work, we describe an approach that is based on a segmentation of common structures of the datasets, which enhances the mutual information and hence, the registration result. An evaluation and a comparison to the registration accuracy of unsegmented datasets is presented.