Mapping volumetric differences between patients with Alzheimer's Disease and normal controls: comparison between univariate analysis and multivariate analysis

• Grant number: 10/17727-2
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: March 01, 2011
• End date: February 28, 2013
• Field of knowledge: Health Sciences - Medicine - Psychiatry
• Principal Investigator: Geraldo Busatto Filho
• Grantee: Luciano de Menezes Sanchez
• Host Institution: Faculdade de Medicina (FM). Universidade de São Paulo (USP). São Paulo , SP, Brazil

Abstract

In the last decades, there have been many advances in the understanding of the neuropathology, neuropsychology and more recently in the genetics of Alzheimer's disease (AD), the most common form of dementia. Although it is known that the main histopathological findings are deposition of senile plaques and neurofibrillary tangles, there are no biological markers able to accurately diagnose AD in preclinical stages. Studies comparing structural magnetic resonance imaging (sMRI) in patients with AD and normal controls were able, in recent years, to detect patterns of brain atrophy confirming previous neuropathological findings and reveal areas of neuronal loss not previously evident. The technique of voxel based morphometry (VBM), for example, is an automated segmentation method of brain structures that allows the comparison voxel-by-voxel, using univariate analysis, changes between groups of images of individuals. However, VBM methods have limitations when dealing with complex, multivariate and dynamic patterns of atrophy. An alternative is to use the methodology of multivariate principal component analysis (PCA) associated with the maximum linear discriminant analysis (MLDA), which allows comparisons between images of individuals considering inter-regional differences and also to perform efficient classification of individual cases. The project goal is to compare the performance of the VBM method and the multivariate method for volumetric changes of the whole brain between cognitively normal controls and patients with AD, and also assess the accuracy of classification into two groups made by the multivariate method using a clinical sign, the neuropsychological test Mini Mental State Examination (MMSE). The candidate will use a database available for download for the entire global scientific community called Alzheimer's Disease Neuroimaging Initiative (ADNI), composed of images of sMRI of 200 control subjects and 200 patients with AD, along with numerous clinical variables, in which the candidate for this scholarship is already registered. (AU)