Automatic volumetric registration of NCCT and CTA brain images using intensity based image registration

In this paper we present a pre-processing stage for an automated volumetric CT stroke image diagnosis system. It concerns the automatic intensity-based 3D image registration of non-contrast CT (NCCT) to CT angiography (CTA) images. The NCCT were acquired with a gantry tilt and hence they are first tilt corrected. After the tilt correction process, the NCCT images are then registered to the CTA image. The accuracy of image registration is measured by the overlap of the segmented skull from both images. The limitation of the current implementation is computing time which should be improved for a clinical application of the method.