Use of the Higuchi's fractal dimension for the analysis of MEG recordings from Alzheimer's disease patients

• Published in: Medical engineering & physics Vol. 31; no. 3; pp. 306 - 313
• Main Authors: Gómez, Carlos, Mediavilla, Ángela, Hornero, Roberto, Abásolo, Daniel, Fernández, Alberto
• Format: Journal Article
• Language: English
• Published: England Elsevier Ltd 01.04.2009
• Subjects: Aged; Aged, 80 and over; Algorithms; Alzheimer Disease - diagnosis; Alzheimer Disease - physiopathology; Alzheimer's disease; Automatic Data Processing; Diagnosis, Computer-Assisted - methods; Female; Fractal dimension; Fractals; Higuchi's algorithm; Humans; Magnetoencephalogram; Magnetoencephalography - methods; Male; Middle Aged; Models, Statistical; Neural Networks (Computer); Radiology; Signal Processing, Computer-Assisted; Stationarity; Surrogate data; Higuchi's algorithm; Stationarity; Fractal dimension; Magnetoencephalogram; Surrogate data; Alzheimer's disease
• ISSN: 1350-4533; 1873-4030
• DOI: 10.1016/j.medengphy.2008.06.010

Alzheimer's disease (AD) is an irreversible brain disorder of unknown aetiology that gradually destroys brain cells and represents the most prevalent form of dementia in western countries. The main aim of this study was to analyse the magnetoencephalogram (MEG) background activity from 20 AD patients and 21 elderly control subjects using Higuchi's fractal dimension (HFD). This non-linear measure can be used to estimate the dimensional complexity of biomedical time series. Before the analysis with HFD, the stationarity and the non-linear structure of the signals were proved. Our results showed that MEG signals from AD patients had lower HFD values than control subjects’ recordings. We found significant differences between both groups at 71 of the 148 MEG channels ( p < 0.01; Student's t-test with Bonferroni's correction). Additionally, five brain regions (anterior, central, left lateral, posterior and right lateral) were analysed by means of receiver operating characteristic curves, using a leave-one-out cross-validation procedure. The highest accuracy (87.8%) was achieved when the mean HFD over all channels was analysed. To sum up, our results suggest that spontaneous MEG rhythms are less complex in AD patients than in healthy control subjects, hence indicating an abnormal type of dynamics in AD.