Neural network analysis of diffusion-tensor MRI data to determine the dominant pathology of the brain

In this paper, neural network analysis of diffusion-tensor magnetic resonance imaging is performed to identify the most informative brain structures for determining the dominant pathology in cases of suspected cerebral microangiopathy or Alzheimer's disease. The data obtained for 19 regions of the brain are studied. They are pre-processed and visualized using Kohonen self-organizing maps. A number of applicant areas for the classifier construction are highlighted. Additional verification to confirm the result obtained is carried out using a multilayer perceptron.