This thesis investigates MRI analysis tools with the potential to highlight structural changes in the brain which occur as a result of neurodegeneration. Traditional MR volumetry was performed to identify regional volume changes associated with healthy aging, Alzheimer's disease (AD) and fronto-temporal lobar degeneration (FTLD). This approach was found to be effective at identifying volumetric differences between groups, even in the presymptomatic stages of AD. Characteristic patterns of atrophy were identified in AD and FTLD groups, which has application in the differential diagnosis of dementia. Techniques for automated analysis of MR images were then examined. A novel tool for the automated propagation of hippocampal regions on serial imaging using fluid registration is described. This was found to produce highly reproducible results, greatly reducing operator time compared to manual delineation of the hippocampus. A detailed comparison of voxel-based morphometry (VBM) and volumetric analysis was performed, which highlighted areas of inconsistency between the two techniques. Examination of the VBM pre-processing steps identified inconsistencies and biases which may be introduced by this process, which may have an impact on resulting statistical parametric maps (SPMs). SPM analysis of fluidly-registered serial images was applied to examine changes in individuals with sporadic AD compared with controls, revealing regions of significant difference consistent with volumetric and pathological data. The same technique was performed on controls and groups of AD subjects at different stages of the disease. The pattern of regional atrophy was found to alter with disease progression. The application of these techniques to the diagnosis and understanding of Alzheimer's disease is discussed. In addition, the implications for the monitoring of disease progression, and the assessment of drug efficacy are considered.