The primary goal of this thesis is to develop an automatic preprocessing step for isolating the brain and correcting RF inhomogeneity in clinical MRI scans. This is necessary for all automatic and semiautomatic MS lesion segmentation algorithms. To this end, we designed, implemented and evaluated a novel, fully automatic intracranial boundary detection and RF correction scheme.
Our intracranial boundary detection technique can also be used by MRI registration algorithms and by Anderson's MRI compression algorithm which requires a completely unsupervised detection technique so that user interaction is not required to store or transmit data.