Reducing The Stimulation Threshold And Energy Consumption Of Retinal Ganglion Cell With Stochastic Resonance

ObjectiveEpiretinal prosthesis is a kind of artificial organ that restores functional vision to the blind by electrically stimulating surviving retinal neurons with an electrode array implanted on top of the retina.Its target cells are mainly retinal ganglion cells(RGCs).Epiretinal prosthesis is widely applied to scientific research and clinical application and has been commercialized,but it needs to be improved in the accuracy of image processing,stimulus electrode,visual resolution and accuracy,the biocompatibility and life of device.The microelectrode array of epiretinal prosthesis is implanted in the retina for a long time,so energy consumption is a very important aspect that affects prosthetic performance.High stimulation threshold can produce excessive energy consumption which may cause tissue damage and affect the life of the prosthesis.There are many ways to reduce the stimulation threshold and energy consumption of epiretinal prosthesis,but most of them are in the research stage.Biphasic rectangular charge-balanced pulses are the classical stimulus waveforms applied to epiretinal prosthesis.Here,a method for reducing stimulation threshold of epiretinal prosthesis without changing the biphasic rectangular pulse is expected to be found.MethodsStochastic resonance(SR)can be used to improve the detection and information transmission of neurons to sub-threshold signals.It has been studied and applied in cochlear implant,ventilator and other prosthetic devices.Therefore,this paper explored SR phenomenon in RGC,the influence of SR on the detection and information transmission of RGC to sub-threshold signal and applying the characteristics of SR to epiretinal prosthesis stimulation.In this paper,a multi-compartment RGC model of extracellular stimulation simulated that epiretinal prosthesis stimulates RGC.The Fohlmeister-Colman-Miller(FCM)model was used to describe the membrane properties of RGC.The extracellular stimulation electrode was an ideal single-phase point electrode,acting on the RGC axon.The stimulation of model was a sub-threshold signal with noise.The sub-threshold signal was biphasic rectangular pulse sequence,the classical stimulus waveform applied to the prosthesis.The noise signals were stochastic biphasic rectangular pulse sequences,Poisson shot noise similar to the synaptic transmission noise,colored noise ─ Ornstein-Uhlenbeck(OU)process,the traditional Gaussian white noise.The responses of RGC model to stimuli with or without noise were observed.Power norm and mutual information rate were used to evaluate SR phenomenon and the effect of SR under different noise types was compared.The additive noise was companied by additional energy consumption.In order to reduce the adverse effects of noise,the noise only was added to the cathode phase and the anode phase of the stimulus waveform that elicited the RGC response,and it is compared to the energy of the stimulus waveform before adjusting.The parameters of noise generating optimal SR effect were optimized by using Genetic algorithm(GA).Optimal combination of noise parameters with the lowest sum of energy consumption and charges was found and compared with that at the threshold level.ResultsWhen certain intensity of four noises was added to the sub-threshold signal respectively,RGC could generate action potentials.The noise parameters of four noises had an effect on power norm and multi-information rate.When the stochastic biphasic rectangular pulse sequences were added,bigger pulse width caused higher power norm.The RMS amplitudes of stochastic biphasic rectangular pulse sequences required to produce the maximum power norm were within the range of less than 30μA.When Poisson shot noise were added,as the noise correlation rate increased,it is as a whole showed the maximum values of power norm first increased and then reduced.The noise standard deviation required to produce the maximum power norm showed the opposite process.With the noise level increasing,the relationship between the power norms or multi-information rates and noise intensities in four noises as a whole showed the inverted U-like graph.With the noise intensities of four noises increasing,the coherence between the input stimulus and the response of RGC increased,but when the noises further increased,stochastic firing was generated.Compared to four kinds of noises,the coherence between the input stimulus and the response is better under a certain noise intensity of stochastic biphasic rectangular pulse sequences and Poisson shot noise.As the intensity and amplitude of noises increased,longer inter-spike intervals disappeared.In contrast to the RGC responses under four kinds of noises,stochastic biphasic rectangular pulse sequences produced the maximum power norm and multiinformation rate.After adjusting the additional mean,the energy of stimulus waveform decreased,existing the energy lower than that of threshold level.The noise parameters of stochastic biphasic rectangular pulse sequences affected the energy consumption of stimulus waveform.The noise parameters optimization was performed for 8 times independently and the average values of optimal noise parameters were wb =0.1449 ms,Tb =0.4204 ms,[ ]rmsE A =3.8184 ms and the average of minimum cost function was 308.6690 p J.The energy consumption and charge of threshold level were 468.18 pJ,15.30 nC.Compared with them,the sum of energy consumption and charge of the optimized stimulus waveform was reduced by 36.2%.RGC model was stimulated 20 times by the sub-threshold signal with noise generated by applying the average of optimal noise parameters.The average energy and charge of the stimulus waveform were 301.0196 pJ and 11.9644 nC,compared with that of threshold level,they were reduced by 35.7% and 21.8% respectively;The minimum values were 285.4879 pJ,11.6608 nC and they were reduced by 39.0% and 23.8%.ConclusionsNoise has an effect on the response of RGC.Appropriate level of noises added to the sub-threshold signal can enhance the response of RGC.SR exists in the RGC and stochastic biphasic rectangular pulse sequences,Poisson shot noise,OU process and Gaussian white noise all can generate and adjust SR phenomenon.Compared with other three types of noises,stochastic biphasic rectangular pulse sequences and OU process are easier to generate SR,but the effect of SR generated by the former is better.In addition,it is found that SR can enhance the detection and information transmission of RGC to the sub-threshold signal.Adding the stochastic biphasic rectangular pulse sequences to the cathode and anode phase of the sub-threshold signal helps to reduce stimulation threshold,energy consumption and charge of RGC stimulation.These may be helpful for reducing stimulation threshold,energy consumption and charge of epiretinal prosthesis and tissue damage,applying SR to epiretinal prosthesis stimulation and improving the performance of epiretinal prosthesis.