Automatical Adaptation of Anatomical Masks to the Neocortex

We describe an image processing chain that is capable of identifying sulci and gyri in MRI brain slices. Contrary to current interactive map fitting schemes it tries to simulate a radiologist’s way of image analysis — a process we call image understanding by landmark detection. In a nutshell, we detect the entry points of the neocortical sulci by an automated procedure. These entry points are identified as belonging to a specific sulcus by comparison with an anatomical database. From these landmarks a further analysis of the surrounding region can be performed. This algorithm is used for an anatomical mapping facility in a multimodal image editor for medical volume datasets.