Electrophysiological Assessment of Internal Noise in the Human Visual Pathway

Purpose. Noise is present throughout the visual system, from photoreceptors to visual cortex. The internal noise of the visual system that limits function has been studied for several decades using behavioral techniques in humans, which is an inherently subjective approach. Furthermore, behavioral noise measurements represent the combined contributions from all sites within the visual pathway, such that retinal noise cannot be separated from cortical noise. The goal of this thesis is to develop objective, electrophysiological methodologies to estimate noise that arises from different sites within the visual pathway. This was accomplished by completing the following three Specific Aims: Aim 1 developed novel noise-based electrophysiological measures. Aim 2 determined how stimulus temporal frequency affects internal noise measurements. Aim 3 developed a simplified protocol that can be applied to patient populations to study how pathology affects internal noise. Methods. Five control subjects and two subjects with diabetic retinopathy were recruited. Amplitude of the flicker electroretinogram (fERG; a measure of photoreceptor and bipolar cell function), the pattern electroretinogram (pERG; a measure of retinal ganglion cell function), and the flicker visual evoked potential (fVEP; a measure of cortical function) were measured as a function of stimulus contrast. Amplitude of the fundamental and second harmonic responses were derived by Fourier transforms. Measurements were performed in the absence of luminance noise and in white luminance noise of different power. Threshold, defined as the minimum stimulus contrast needed to elicit a measurable response, was derived from the amplitude measures based on Naka-Rushton fits to the response amplitude vs signal contrast data. Threshold was then plotted as a function of luminance noise power and the data were fit with the Linear Amplifier Model, a common model of visual performance in noise. A clinically optimized protocol was developed and implemented based on the data of Aims 1 and 2. Results. Luminance noise had no effect on the fundamental component of the fERG and fVEP. Consequently, fundamental fERG and fVEP contrast thresholds were independent of noise power. However, noise did reduce the second harmonic component of the fERG, fVEP, and pERG. The mean internal noise (Neq) for the fERG (0.38±0.04) was greater than that of the fVEP (0.17±0.03), but similar to that of the pERG (0.33±0.05). Stimulus temporal frequency had no effect on Neq for the fERG and slightly increased Neq with increasing temporal frequency for the fVEP. Contrast threshold and Neq in diabetic retinopathy were normal for the fERG and fVEP, but elevated for the pERG. Conclusion. This thesis provides the first objective assessment of internal noise in the human visual pathway using electrophysiology. The effect of noise on the second harmonic, but not the fundamental, for each measure can be accounted for by a linear-nonlinear-linear cascade model. The surprising finding that cortical noise is lower than retinal noise can likely be attributed to cortical spatiotemporal summation, which could reduce noise in the fVEP. Finally, data from a small sample of diabetic subject provides proof-of-concept that the approach developed in this thesis has potential clinical utility.