Abstract

Alzheimer’s disease (AD) is a progressive disorder that affects cognitive brain functions and develops many years before there are any clinical manifestations. A biomarker that provides a quantitative measure of changes in the brain in the early stages of AD would therefore be useful for early diagnosis. However, this would involve dealing with large numbers of people because up to 50% of dementia sufferers do not receive a formal diagnosis. Thus, there is a need for accurate, low-cost, robust, and easy to use biomarkers that can be used to detect AD in its early stages. Recent guidelines promote the use of biochemical and neuroimaging biomarkers to improve the diagnosis of AD. Cerebral spinal fluid (CSF) testing for AD is not widely used in clinical practice because it involves an invasive lumbar puncture procedure. Neuroimaging (e.g., positron emission tomography-PET), on the other hand, is expensive, available only in specialist centres, and may be unsuitable for patients with pacemakers or certain other implants. Blood-based biomarkers have shown promising results in terms of AD diagnosis, but these are not yet fully developed and low-cost biosensors to detect such biomarkers do not yet exist. However, electroencephalogram (EEG) based biomarkers can potentially fulfil these needs and play a vital role in the early diagnosis of AD. AD causes changes in EEGs that are thought to be associated with functional disconnections among cortical areas due to the death of brain cells. EEG analysis may therefore provide valuable information about brain dynamics in AD. Potentially, the EEG could be used to detect changes in brain signals even in the preclinical stages of the disease. This means it could be used as a first line decision-support tool in AD diagnosis and complement other AD biomarkers. This thesis describes research into the development of EEG biomarkers that detect AD based on analysis of changes in the EEG. The most characteristic features in AD are slowing of the EEG activities, a decrease in coherence, and a reduction in complexity. These changes can be quantified as a biomarker of AD. In this study, we identified characteristic EEG features that have a significant association with AD. The most promising EEG features were then used to develop EEG biomarkers that can exhibit high diagnostic performance. Four measures of complexity were investigated and evaluated for their suitability as the basis for EEG-based biomarkers of AD: Tsallis entropy, Higuchi Fractal dimension, Lempel-Ziv complexity, and approximation entropy. Two EEG slowing measures were also investigated and evaluated: changes in zero-crossing intervals, and changes in the power spectrum of EEG. In addition, a new approach to quantifying the slowing of EEGs based on analysing changes in EEG amplitudes was developed and evaluated. The coherence of connections among cortical regions of the brain was also investigated to analyse EEG connectivity. A new biomarker was developed based on analysing changes in EEG amplitude (ΔEEGA). This is a marker for the subsequent rate of cognitive and functional decline in AD patients and provides high diagnostic performance. The performance of ΔEEGA measured 100% and 88.88% for sensitivity and specificity, respectively. Our results therefore show that EEG-based measures can potentially be a good biomarker for AD. An important contribution of the thesis is the development of a method to derive robust biomarkers from the EEG through selective band filtering and by combining key biomarkers. Thus, this study provides a framework for constructing robust EEG biomarkers that can be used to detect AD with high diagnostic performance (e.g., in terms of sensitivity and specificity).

Keywords

Alzheimer’s Disease, Alzheimer’s Disease Detection, EEG Biomarkers, EEG complexity, EEG Slowing, EEG Coherence

Document Type

Thesis

Publication Date

2019

Download

Share

COinS