Alzheimer's Disease Diagnosis Based on Cognitive Methods in Virtual Environments and Emotions Analysis

Abstract

Dementia is a syndrome characterised by the decline of different cognitive abilities. Alzheimer's Disease (AD) is the most common dementia affecting cognitive domains such as memory and learning, perceptual-motion or executive function. High rate of deaths and high cost for detection, treatments and patient's care count amongst its consequences. Early detection of AD is considered of high importance for improving the quality of life of patients and their families. The aim of this thesis is to introduce novel non-invasive early diagnosis methods in order to speed the diagnosis, reduce the associated costs and make them widely accessible. Novel AD's screening tests based on virtual environments using new immersive technologies combined with advanced Human Computer Interaction (HCI) systems are introduced. Four tests demonstrate the wide range of screening mechanisms based on cognitive domain impairments that can be designed using virtual environments. The use of emotion recognition to analyse AD symptoms has been also proposed. A novel multimodal dataset was specifically created to remark the autobiographical memory deficits of AD patients. Data from this dataset is used to introduce novel descriptors for Electroencephalogram (EEG) and facial images data. EEG features are based on quaternions in order to keep the correlation information between sensors, whereas, for facial expression recognition, a preprocessing method for motion magnification and descriptors based on origami crease pattern algorithm are proposed to enhance facial micro-expressions. These features have been proved on classifiers such as SVM and Adaboost for the classification of reactions to autobiographical stimuli such as long and short term memories.

Abstract: arXiv descriptive metadata (CC0). Thank you to arXiv for use of its open access interoperability. This service was not reviewed or approved by, nor does it necessarily express or reflect the policies or opinions of, arXiv.