Quadratic Sample Entropy and Multiscale Quadratic Sample Entropy of the Electroencephalogram in Alzheimer’s Disease

Introduction Non-linear analysis techniques have been reported as a way of characterizing Alzheimer’s disease (AD) from electroencephalograms (EEG), which could be utilised for early and more accurate detection of the disease. Among them, entropy has been used increasingly for EEG studies since the introduction of Approximate Entropy (ApEn) and Sample Entropy (SampEn). A newly developed entropy measure, Quadratic Sample Entropy (QSE), overcomes some limitations of these methods by ‘normalizing the match count to the volume of the matching region’ [1]. Another technique, Multiscale Entropy (MSE) [2] aims to calculate the SampEn of data over a range of time scales by coarse graining time series i.e. the EEG before calculating the entropy, giving more information on the brain complexity. In this pilot study we have used QSE and a novel approach of combining MSE with QSE (MSEQSE) to identify changes in electrical activity in the brain caused by AD.