Purchase this article with an account.
S. K. Gardiner, W. H. Swanson, S. Demirel, A. Turpin, A. M. McKendrick, C. A. Johnson; Examining the Perimetric Sensitivity-Variability Relation in Glaucoma Using a Cortical Spike Model. Invest. Ophthalmol. Vis. Sci. 2007;48(13):1617.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
It is well-established that the variability of sensitivity estimates from standard automated perimetry tends to increase in glaucoma subjects as sensitivity decreases. The reasons for this relation are not yet clear. This project aims to produce a model of perimetric stimulus detection in glaucomatous eyes which is capable of explaining the relation.
In the model, individual retinal ganglion cells (RGCs) produce neural spikes at a rate determined by the stimulus luminance. At the second stage, these spikes combine with noise from other cortical cells to produce the sum of excitatory postsynaptic potentials (EPSPs) of a given cell in the visual cortex. The cortical cell generates a spike when its sum of EPSPs is greater than some firing threshold, determined for each cell by the level of background noise in the absence of a stimulus. The observer is considered to respond when the largest output from these cortical channels is sufficiently high. This model can be used to estimate the probability of responding to different luminance stimuli, and so generate a frequency-of-seeing (FOS) curve. The perimetric sensitivity (the luminance at which 50% of stimuli generate a response) and the slope of the FOS curve were modeled for healthy eyes, and with different levels of RGC damage.
For the healthy system, with our chosen set of parameters, the mean healthy eye had a sensitivity of 29.7dB (Standard deviation 0.8dB), and the FOS curves had a mean interquartile range (IQR) of 3.6dB (Standard deviation between IQRs 1.0dB), well within the range of published results. For damaged eyes, the IQR increased as sensitivity decreased, reaching as high as 15dB in some cases. The rate of increase of IQR could be systematically changed by incorporating divergent dysfunction of RGCs into the model, i.e. different RGCs becoming increasingly dysfunctional at differing rates.
This model can generate realistic FOS curves, which become shallower as sensitivity decreases. Cell death and dysfunction both affected variability, and the relation may provide information about the levels of dysfunction in different patients with the same sensitivity loss.
This PDF is available to Subscribers Only