Atlas-Based Fusion of Medical Brain Images Methods and Applications

Abstract: This thesis focuses on the development of methods for fusion of information from medical brain scans. The concept of medical image fusion refers to the process of extracting and utilising information from several scans simultaneously in the analysis and diagnosis of patients.One very important part of the fusion process is the image registration, which is used to find a mapping or transformation of points from one image to the corresponding points in another image. This can, for example, be used to correct for relative movements between patient examinations, thus, making direct comparisons between different scans possible. Furthermore, the registration can be used to map images from different individuals into a common standard anatomy. This is important, since it enables comparisons between the individuals and also between whole groups of individuals. In the thesis, both methods to be used for registration between scans from the same individual and for scans from different individuals are presented.Another part of the thesis is directed towards analysis of brain scans. Most of the methods are based on a computerised brain atlas, which defines a standardised mapping of the brain into sub-regions. These regions are either anatomical or functional and can be used for a more detailed analysis of the brain scan. The presented methods cover general methods for comparisons of single patients with groups of individuals, methods for feature calculations from brain atlas defined regions, and methods for extraction of more advanced features for automatic classification of brain scans.Furthermore, image visualisation is always an important part in medical imaging. This is because the constantly increasing amount of medical information demands more advanced visualisation techniques to enhance and aid the interpretation of the data. The methods presented in this thesis are focused on combined visualisation of multiple brain scans, which is useful when scans expressing different types of information are available. For instance, a combined visualisation can be helpful to detect anatomical regions of specific functional importance in the brain.