Automatic Localization of the Hippocampal Region in MR Images to Asses Early Diagnosis of Alzheimer’s Disease in MCI Patients

Atrophy and other brain changes, which are typical of aging, generate wide inter-individual variability of morphology in the medial temporal lobe (MTL), including the hippocampal formation. Starting from a sample population of 133 MR images we developed a procedure that extracts from each MR two sub images, containing the hippocampal formations plus a portion of the adjacent tissues and cavities. Then, a small number of templates is selected among the previously obtained sub images, able to describe the morphological variability present in the whole population. Finally an automatic procedure is prepared which, on the basis of the given set of templates, is able to find both hippocampal formations in any new MR image. MR images ranging from normalcy to extreme atrophy can be successfully processed. The proposed approach, besides being a preliminary step towards the unsupervised segmentation of the hippocampus, extracts from the MR image information useful for diagnostic purposes and, in particular, could give the possibility of performing morphometric studies on the medial temporal lobe in an automated way. The automated analysis of MTL atrophy in the segmented volume is readily applied to the early assessment of Alzheimer Disease (AD), leading to discriminating converters from Mild Cognitive Impairment (MCI) to AD with an average three years follow-up. This procedure can quickly and reliably provide additional information in early diagnosis of AD.