Simulation of surgical procedures in computer based simulator systems is a relatively new area of interest in the field of medical technology. Many different systems are under development, but it is still far until this technique becomes routine in a clinical environment.

One limitation with these systems is that the computer generated models of the anatomy are ususally not patient specific. In general these systems contain one or a couple of generic models, reflecting the normal anatomy. In order to use these systems as tools for preoperative planning, it must be possible to generate computer models from real patient cases. This is relevant also when these systems are used for training and education, since the surgeon can practice on a variety of real cases, ranging from routine procedures to more complex, or more unusual, tasks.

To produce patient specific models a segmentation process is generally involved. A manual segmentation method, where a user delineates the interesting structure slice by slice in a patient dataset, is a very tedious process, and the result tends to vary between different users. The results would be more consistent, and the process would be more time and cost efficient, if a computer performed this task automatically, with a limited amount of user interaction.

This work focus on automatic segmentation of hip bones from computer tomography scans, in order to generate patient specific computer models for a hip fracture surgery simulator. The segmentation is carried out by matching a template dataset to a patient dataset using a method called the morphon method. Each point in the template is categorised as one of three tissue types, soft tissue (background), spongy bone (bone interior), or cortical bone (bone surface). When this dataset has been matched to the patient dataset the information in the template can be projected onto the patient dataset and used to find the different objects in this dataset.

The presented method can be used in other applications as well. One example demonstrated in this thesis is segmentation of a structure in the brain called the hippocampus.