The visual evoked potential (VEP) reflects the electrical activity in the cerebral cortex due to visual stimulation. It contains information on signal processing in the visual pathways from the retina, through the lateral geniculate nucleus (LGN) to the cortical level. Although the VEP is generated by various sources in the multi-layer neural network, the functions of some neuronal mechanisms and the transfer characteristics of particular pathways can be revealed and analyzed by use of carefully designed stimuli (Zemon & Ratliff, 1984; Zemon, Victor & Ratliff, 1986). This dissertation focuses on studies of the visual evoked potential as a composite of neuronal activities such as direct-through excitation, local lateral inhibition and contrast gain control.
The work concentrates on the sandwich model, a three stage combination of linear-nonlinear-linear elements, introduced in 1970 by Spekreijse & Oosting, to represent signal flow in the visual pathways. Previous efforts to identify the three elements involved stimulation with contrast reversing spatial patterns using a two-sinusoid temporal signal (Spekreijse & Reits, 1982; Zemon & Ratliff, 1984; Zemon, Victor & Ratliff, 1986). Results of these studies are limited to amplitude and phase characteristics of the two linear elements in terms of sum and difference frequency components of the VEP.
In the current work, transfer functions were sought for the two linear elements in the original three-stage sandwich system in order to obtain an analytic description of the system. The goal was to represent the frequency responses, including those for sum and difference frequencies obtained from two-sinusoid stimulation, as well as transient responses elicited by step (square-wave) contrast reversals. Data from ten normal subjects were analyzed. To fit the observed data collected from those subjects, it was found that the first linear element in the sandwich system must be a non-minimum phase function with zeros in the right half s-plane.
Based on prior investigations of single-cell responses in the cat retina (Shapley & Victor, 1978; 1981; Victor, 1981) and the VEPs in humans studied with a two-sinusoid contrast reversing pattern (Zemon, Victor & Ratliff, 1986; Zemon, Conte & Camisa, 1987), it appears that the amplitude-phase relation of the VEP frequency response to two-sinusoid stimulation depends on two inhibitory mechanisms, contrast gain control and lateral interaction. The phenomena of the inhibitory processes in the VEP were demonstrated in the current work by tests that included three-sinusoid stimulation, which enabled the investigation of direct-through excitatory and lateral inhibitory interactions simultaneously.
A new model has been proposed based on the sandwich system with emphasis on its physiological interpretation. This extended model incorporates contrast gain control and lateral inhibitory mechanisms in an inhibitory parallel path, which permits the analysis of separate excitatory and inhibitory processes. The effects of these inhibitory mechanisms are represented in terms of parameter control in the basic sandwich model. System identification procedures have been developed, and model parameter estimation and validation were performed for two individual subjects.
In the first modified model, the parameter control has been designed to represent steady-state operation. This model provides a good fit for the VEP frequency responses corresponding to the two-sinusoid and single sinusoid stimulation, but fails to represent the VEP transient response. Further modifications resulted in a second model incorporating dynamic parameter control. The second model provides a good fit for the VEP transient response; at the expense of somewhat poorer fit of the frequency responses.
|
![[NJIT logo]](http://archives.njit.edu/vhlib/images/njit_logo.gif){kind=logo}
![[Van Houten Library logo]](images/vhl-top-image.jpg)
