UNIVERSITY OF BUCHAREST
FACULTY OF PHYSICS

Guest
2024-11-22 2:12

 HOME     CONFERENCES     SEARCH            LOGIN     NEW USER     IMAGES   


Conference: Bucharest University Faculty of Physics 2010 Meeting


Section: Biophysics and Medical Physics;Electricity and Magnetism


Title:
Computational analysis of non-linear dynamical component of the eye electric activity


Authors:
Eduard Dorobantu, Lisette Oancea, D. Creanga, V. Tura


Affiliation:
Univ. Al. I. Cuza Iasi, Faculty of Physics, 11 A Blvd. Carol I, Iasi


E-mail
mdor@uaic.ro


Keywords:
biological signal, non-linear analysis, chaos theory


Abstract:
The visual system electric activity can be investigated by means of the electrographic signal recorded at the level of the cornea. It is generated by various photoreceptor cells as well as by neural cells from the retina structure. The study presented in this paper is dedicated to the analysis of the electroretinographic signal recorded for intermittent light illumination with a frequency of about 10 Hz. Though the periodic component of the visual system would be expected to be the dominant however certain fluctuations could be noticed so that their weight in the system dynamics became the focus of the investigation. Correlation dimension and Lyapunov largest exponent have been calculated by using specialized software for three groups of data series; first group represent the raw electroretinographic signal recorded for three different levels of light intensity; the second and the third group represent the smoothed signals – once and respectively twice. Both computational parameters chosen for the analysis were graphically plotted against the embedding dimension. Comparative discussion was attempted considering the role of the numerical smoothing of the signal versus the possible loss of intrinsic information from the investigated signal. The non-linear dynamical component was revealed according to chaos theory, taking into account the magnitude and the sign of the Lyapunov exponent as well as the non-integer values toward the correlation dimension tends to saturate.