EEG amplitude modulation analysis for semi-automated diagnosis of Alzheimer’s disease

Abstract

Recent experimental evidence has suggested a neuromodulatory deficit in Alzheimer's disease (AD). In this paper, we present a new electroencephalogram (EEG) based metric to quantitatively characterize neuromodulatory activity. More specifically, the short-term EEG amplitude modulation rate-of-change (i.e., modulation frequency) is computed for five EEG subband signals. To test the performance of the proposed metric, a classification task was performed on a database of 32 participants partitioned into three groups of approximately equal size: healthy controls, patients diagnosed with mild AD, and those with moderate-to-severe AD. To gauge the benefits of the proposed metric, performance results were compared with those obtained using EEG spectral peak parameters which were recently shown to outperform other conventional EEG measures. Using a simple feature selection algorithm based on area-under-the-curve maximization and a support vector machine classifier, the proposed parameters resulted in accuracy gains, relative to spectral peak parameters, of 21.3% when discriminating between the three groups and by 50% when mild and moderate-to-severe groups were merged into one. The preliminary findings reported herein provide promising insights that automated tools may be developed to assist physicians in very early diagnosis of AD as well as provide researchers with a tool to automatically characterize cross-frequency interactions and their changes with disease.

Towards automated EEG-Based Alzheimer's disease diagnosis using relevance vector machines

“…These features have been used as indicators of cortical connectivity [31]. As we are interested inter-hemispheric connectivity, coherence features (both magnitude and phase) were computed only for the 8 Lastly, amplitude modulation features have been recently proposed and shown to reliably detect AD, as well as monitor disease progression [5], [32], [33]. To compute the features, the EEG full-band signal is first decomposed into the five conventional sub-bands: delta, theta, alpha, beta, and gamma.…”

“…Since artefacts can have detrimental effects on EEG-based AD diagnostics, the majority of the published works have utilized artefact-free EEG segments (called epochs) which have been selected by expert clinicians through exhaustive visual inspection. This manual selection process introduces additional biases and errors into the diagnostic procedure [17], as well as renders it "semi-automated" [5], thus still making it costly and time-consuming. As an alternative, automated artefact removal (AAR) algorithms may be used to remove artefacts from the EEG signal without the need for human intervention.…”

“…Several different effects of AD on the EEG signal have been reported in literature. The most distinctive of these effects include: i) noticeable shift occurring in the EEG power spectrum (e.g., slowing of the EEG) [8]- [10]; ii) reduced inter-hemispherical spectral coherence [11]- [13]; iii) decreased EEG "complexity," likely caused by the reduction in nonlinear connections between cortical regions or even neuronal death [7]; and more recently, iv) a change in neuromodulatory behaviour measured via EEG amplitude modulation analysis [5], [14]. Many of these changes have been shown to be related [15] and diagnostic accuracies in line with more advanced neuroimaging techniques (e.g., fMRI) have been achieved [16].…”

“…To overcome these abovementioned limitations, quantitative electroencephalography (qEEG, henceforth referred to as "EEG") has been proposed as a promising tool to assist physicians in the diagnosis of AD [5]- [7]. The EEG signal is a record of the change of the electrical field produced by neural activity on the cerebral cortex.…”

“…This paper investigates the benefits of using AAR over manual selection for automated assessment, with the end goal of developing a tool that can be used to aid clinicians (both experienced and inexperienced) during the AD diagnostic procedure. Building on recent evidence that support vector machine (SVM) based classifiers are useful for semi-automated AD diagnostics [5], [21], [22], a secondary aim of this paper is to explore the benefits of using so-called relevance vector machines (RVM) over SVM classifiers. RVMs utilize a probabilistic Bayesian learning framework to find classifiers that have i) fewer basis functions than a comparable SVM and ii) the benefit of explicitly predicting the probability of class membership, thus providing clinicians with a richer pool of information.…”