Registration and analysis of magnetic resonance images of the brain

Registration of magnetic resonance (MR) images with functional magnetic resonance (fMR) images is an essential step in multi-modality brain image analysis. A technique has been developed for registration of three-dimensional MR and fMR images. Using this method, any local shape variations from anatomy in the functional images can be corrected. A qualitative analysis indicating that the new registration technique based on modeling of the voxel in the three dimensional space as a fluid element and using tensor analysis for correction of distortion will be presented and shown to be accurate for registration of MR/fMR correlation studies. When using this method, processing of MR and fMR data is automatic with no operator interaction required. Accurate three-dimensional registration of anatomical and functional brain images can be obtained using the new algorithm without fiducial or external markers.;A labeled anatomical image, when registered with a functional image, allows the quantification of activity in various brain structures such as various cortices. A new method to accurately label the anatomical images depicting the functional information from the registered functional scans has been developed. The methods presented here are demonstrated by co-registering 3D anatomical and functional MR images. However, the method is equally well suited to co-registration across imaging modalities; for example, metabolic images from Positron Emission Tomography can be registered to anatomical images from MRI or Computed Tomography. The use of registration algorithms on the anatomical and functional images and identification of functional information in the functional scans are described in this dissertation.