CHAOTIC QUANTIFIERS OF EEG-SIGNAL FOR ASSESSING PHOTO-AND CHEMOTHERAPY

Analysis of EEG-signal using methods of nonlinear dynamics and deterministic chaos theory may supply clinicians with effective quantitative descriptors of underlying nonlinear dynamics and chaos in the brain, so helping in neuro-psychiatric diagnosis and in assessing applied therapy. Howeve r, calculation of „standard” chaotic quantifiers requires long stationary EEG-signal epochs and needs relatively long processing time. So new quantifiers are needed. Introduction Being a highly complex nonlinear system brain shows features characteristic for deterministic chaos [1]. So, it is not surprised that methods of nonlinear dynamics and chaos theory may be more appropriate than linear methods for neuro-psychiatric diagnosis and for assessing influence of applied therapy through analysis of EEG-signal, which represents overall electric activity of the brain,. Materials and Methods For EEG-data collection we used computerised 16 channels P.I.M. ELMIKO (Warsaw) DigiTrackTM system. Sampling frequency was 128, 240 or 256 Hz. Epochs for analysis were extracted from standard EEG-records. We analysed EEG-signals of healthy (voluntary) subjects, under drug-free condition and under the influence of Diazepam. (Valium) and EEG of subjects with Seasonal Affective Disorder (SAD) before phototherapy (BLT) and two weeks after BLT treatment. We tried to use „standard” chaotic quantifiers for EEG-signal analysis. We used simultaneous coordinates (multichannel) method for phase-space reconstruction, and Karhunen-Loeve transform and cumulative pattern entropy for complexity analysis; we also tried attractor reconstruction using single-channel EEG-data by timedelay method and we computed such chaotic quantifiers as time delay, embedding dimension, pointwise correlation dimension, the largest Lyapunov exponent. Results Unfortunately, the results obtained show no consistent pattern of changes in standard chaotic quantifiers when EEG-signals before and after therapy are compared (Fig. 1; cf. also [2,3]). So new quantifiers are needed. Fractal dimension of EEG-signal, calculated using Higuchi’s algorithm [4-6], seems to be the most promising of such quantifiers (Fig. 2). t S P a t i e n t 6 2 3 9 8 7 6 5 4 3 2 1 0 1 1 0 0 8 0 6 0 4 0 2 0 0 t S P a t i e n t 6 3 8 8 7