A new technique for fractal analysis applied to human, intracerebrally recorded, ictal electroencephalographic signals

Application of a new method of fractal analysis to human, intracerebrally recorded, ictal electroencephalographic (EEG) signals is reported. 'Frameshift-Richardson' (FR) analysis involves estimation of fractal dimension (1 < FD < 2) of consecutive, overlapping 10-s epochs of digitised EEG data; it is suggested that this technique offers significant operational advantages over use of algorithms for FD estimation requiring preliminary reconstruction of EEG data in phase space. FR analysis was found to reduce substantially the volume of EEG data, without loss of diagnostically important information concerning onset, propagation and evolution of ictal EEG discharges. Arrhythmic EEG events were correlated with relatively increased FD; rhythmic EEG events with relatively decreased FD. It is proposed that development of this method may lead to: (i) enhanced definition and localisation of initial ictal changes in the EEG presumed due to multi-unit activity; and (ii) synoptic visualisation of long periods of EEG data.