A novel method for analysis of EEG background activity in epileptic patients and healthy subjects using Hilbert transform

Hypothesising that non-linear analysis of Electroencephalogram (EEG) signals may provide clinicians with information for medical diagnosis and assessment of the applied therapy, we investigated the EEG background activity in normal and epileptic subjects using Hilbert Transform Scatter Plots (HTSPs). A visual inspection of HTSPs revealed distinct differences in the different physio-pathological states of the EEG. These differences were quantified employing Radial Distance Index (RDI), which is derived from the Central Tendency Measure (CTM) of the HTSP. The application of such a technique is justified by ascertaining the presence of non-linearity in the EEG time series through surrogate test. The false-positive rejection of the null hypothesis is eliminated by employing Welch window before the computation of the Fourier transform and randomising the phases, in the generation of the surrogate data. Receiver Operating Characteristic (ROC) analysis reveals that RDI can discriminate between seizure and non-seizure states with very high accuracy.