Classification of brain MR image data using data mining techniques

Alzheimer's Disease is a disease that shows its effects with the progression of the age and it causes the brain to be unable to fulfill its expected functions. The disease's effects show variety according to its phase such as to forget the name of surrounding people or cannot continue daily life without help. As of we know, there is no method has general acceptance for diagnosis and treatment. In this thesis study, the performance of data mining methods and that of deep learning on diagnosis of Alzheimer's Disease (AD) is assessed by using patients' magnetic resonance image (MRI) data. MRI data from 144 healthy persons named control normal (CN) and 175 patients with AD are used. These two groups consist of 167 male and 152 female aged between 55 and 91. Volumetry statistics of the parts of the brain from related MRI data are obtained and performance of the traditional data mining algorithms on feature selection and classification are evaluated. Also, the performance of Artificial Bee Colony algorithm developed in this thesis and used as a feature selector is compared with the classical feature selector methods. Besides this, collected images segmented as Gray Matter (GM), White Matter (WM) and Cerebrospinal Fluid (CSF) after the normalisation process. Statistical Parametric Mapping (SPM) software's tools are utilized for the normalization and segmentation. These GM images are used in deep learning method. Results are compared.