Prospects for routine detection of dementia using the fractal dimension of the human electroencephalogram

Reports our initial study to demonstrate that subject-specific variability of the fractal dimension of the human electroencephalogram (EEG) is an important candidate method for the routine detection of dementia. The results in this paper show that the adapted box dimension of the auto-correlation function and the dimension of the zero-set for the auto-correlation function are good candidates for use in the subject-specific detection of dementia. This is because the measured fractal dimension of the EEG is generally lower for a demented subject than for normal subjects, and the variability of a single normal subject´s fractal dimension is small in comparison to the variability between members of the set of normals. The results indicate that this method may have the potential to achieve our goal of automated EEG analysis, providing an opportunity to maximise the quality of life for patients by using modern treatments