3D characterization of Hippocampus in Dementia

Dementia is the greatest global health challenge of our generation, and its prevalence has increased over the past few decades. Approximately 44million people across the world living with it. There is currently no "cure" for dementia although neuroimaging which is progressively regarded as an essential and important tool in the diagnostic workup of the study of cerebral pathologies, may be also helpful for developing advanced approaches to achieve diagnoses as early as possible for therapies aimed at reducing the development of neurodegenerative diseases. Dementia causes atrophy of the entire brain, although some subcortical regions such as the hippocampus which is one of the essential biomarkers and can be assessed with quantitative volumetric methods. Many works of literature have developed different approaches to characterize dementia and to classifier its stages automatically but there is a bottleneck in the diagnostic performance which was shown in previous methods, due to the missing of efficient strategies for representing neuroimaging biomarkers. In this work, a fully automatic Computer Aided Diagnosis system based on supervised learning methods is proposed to be applied on segmented hippocamps and excluded a mass lesion from a 3D MRI OASIS database which includes a cross-section of 175 subjects to predict dementia stages. Our result evinces that the proposed approach can be used to characterize and predict dementia with over 96% accuracy, considerably higher than that of conventional basic methods. This study confirms that to extract geometric descriptors from the hippocampus can effectively solve the dementia characterization problem.