Retinal prosthetics strive to restore vision to those blinded by outer retinal diseases such as macular degeneration and retinitis pigmentosa. There has been considerable progress in recent years with reports of previously-blind subjects identifying household objects, navigating in limited ways through unfamiliar landscapes and even reading. Despite this progress however, the overall quality of elicited vision is still remais somewhat limited. For example, even the fastest subjects can only read a few simple words per minute and the average reading rate across all subjects is considerably lower. In addition, the resolution from these devices is typically much lower than that predicted by electrode spacing. One of the factors thought to reduce the quality of prosthetic vision is the methods utilized to stimulate retinal neurons. In the healthy retina, approximately a dozen different types of ganglion cells (retinal output neurons) each utilize different signaling patterns to communicate with the brain. For example, ON ganglion cells generate bursts of spiking at the onset of a light stimulus while OFF cells are silent or even reduce spiking (if a non-zero baseline rate is present). In contrast, stimulation from prosthetic electrodes is thought to create highly similar patterns of spiking in many ganglion cells, including both ON and OFF ganglion cells simultaneously and thus transmit a signal to the brain that is non-physiological. Recently, we tested a series of amplitude-modulated waveforms: 2000 pulse per second (PPS) constant-amplitude train with an occasional increase (or decrease) in amplitude, i.e. an increase from 50 uA (baseline) to 60 uA over the course of 150 ms followed by a return to 40 uA over the subsequent 150 ms. As expected, such waveforms elicited bursts of spikes in ON BT cells for each occurrence of the transient increase. Surprisingly however, responses in OFF BT cells were quite different and consisted of a reduction in spiking during the transient increase in stimulus amplitude. Thus the same stimulus waveform elicits an increase in spiking in ON brisk transient (BT) cells and a simultaneous decrease in spiking in OFF BT cells. This closely matches the physiological response pattern for these two cell types raising the possibility that this approach may have advantages over existing stimulation methods. Our goal in this proposal is to investigate these differences further by exploring their sensitivity to the parameters of stimulation with the goal of optimizing the underlying stimulation process. Additional preliminary experiments indicate that the response to 2000 PPS originates in the ganglion cell (i.e. it is not mediated by the synaptic circuitry). Therefore, we hypothesize that the response differences arise from intrinsic differences across ganglion cell types probably differences within the axon initial segment (AIS). Therefore, we will study the AIS differences across types in order to develop accurate computational models that can be used to understand and hopefully further enhance the response differences. Finally, we will also study how both responses as well as the underlying biophysical features change as the retina degenerates.

Public Health Relevance

Retinal prosthetics can restore some elements of vision to those blinded from retinal diseases but overall quality is limited. To improve outcomes, we are developing stimulation schemes that bring the neural activity elicited artificially more in line wih normal physiological patterns. Here, we have found a way to replicate the differences in ON vs. OFF ganglion cells - a key component of normal retinal signaling.

National Institute of Health (NIH)
National Eye Institute (NEI)
Research Project (R01)
Project #
Application #
Study Section
Special Emphasis Panel (BNVT)
Program Officer
Greenwell, Thomas
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Massachusetts General Hospital
United States
Zip Code