Electrical and Electronic Engineering - Theses

Permanent URI for this collection

http://hdl.handle.net/11343/363

Filters

2016 1
1
Reset filters

Settings
Statistics
Citations
Start date: 2010 × End date: 2019 × Subject: retinal prosthesis × Author: MATURANA, MATIAS ×

Search Results

Now showing 1 - 1 of 1
  • Item
    Thumbnail Image
    A predictive model of retinal ganglion cell responses to electrical stimulation
    MATURANA, MATIAS ( 2016)
    Degenerative diseases such as retinitis pigmentosa and age-related macular degeneration result in the loss of photoreceptor cells, which function to transduce light into a neural signal. However, retinal ganglion cells (RGCs), the output cells of the retina, and other cells within the retina often survive in high numbers. Recent developments in visual prostheses have demonstrated that electrical stimulation of the retina is becoming a viable therapy for those blinded through degenerative diseases. While the developments of retinal prostheses are still in their infancy, clinical trials have shown that the vision produced by retinal prostheses can appear complex in both space and time. The development of improved stimulation strategies for the bionic eye requires an understanding of the effects of RGCs to electrical stimulation. This thesis investigates a model that can be used to predict responses of RGCs to arbitrary patterns of electrical stimulation. Intracellular whole cell patch clamp recordings were made in whole mount preparations from normal sighted rats to develop RGC response models to electrical stimulation. Recordings were made at room temperature (~24ºC). Stimulation was applied using a custom-made multi electrode array and consisted of random amplitude biphasic pulses applied at constant frequency. Short-latency responses were correlated with the stimulation applied and a spike triggered covariance technique was used to determine spatial features of the stimulation that resulted in a response. Generally, the spatial arrangements of electrodes that influenced the cell’s response were as expected: the electrodes closest to the recorded cell had largest influence on the cell’s response. An extracellular recording technique was applied to model the long-latency responses to electrical stimulation. Recordings were made close to physiological temperature (~34ºC). The mathematical model used during the intracellular recordings was adapted to also model temporal features of stimulation. Temporal features of stimulation for many cells were complex; the polarity and spatial organisation of stimulation changed over time. Additionally, both excitatory and suppressive features of electrical stimulation were revealed by the model. The effects of temperature were examined to investigate whether some differences observed in results for the two recording techniques could be explained by the temperature used during the two experiments. In vitro recordings at different temperatures were used to investigate how retinal responses changed at different temperatures. The sensitivity of RGCs to electrical stimulation was found to be higher at temperatures closer to physiological temperature. Additionally, a greater amount of long latency activity was observed, suggesting increased activation of the retinal network. Simulations were used to explore an algorithm for achieving spatial control of neural activation. The algorithm made use of the error between recorded and target responses to fine tune the stimulation applied. The simulations suggested that the model can be used to manipulate spatial interactions in a predictable manner, thereby improving spatial fidelity. Additionally, closed loop stimulation may be used to mitigate undesirable effects of stimulation that are observed clinically, such as fading; a phenomena that results in the visual percept produced by electrical stimulation disappearing over time despite constant stimulation. Electrical stimulation of the retina often results in indiscriminate activation of many RGC types. A major goal of electrical stimulation is the targeted activation of certain cell types, such as ON cells or OFF cells. Traditionally, light responses are used to classify cell types. In the degenerate retina, where light responses are not obtainable, other methods are required to identify cells. Analysis of recorded intracellular responses revealed that the action potential waveform may contain identifiable features that could be used to establish the cell type. A previously developed multi compartment model of a RGC was used to relate features in morphology and electrophysiology to features in the action potential waveform. Overall, the results of my investigations demonstrate that RGC responses to electrical stimulation can be accurately modelled and predicted. Complex spatiotemporal features of electrical stimulation can be extracted and explained in a computationally simple model. The work presented here can aid in future developments of improved stimulation strategies that achieve a tighter control of neural activation.