Abstract
Purpose: :
Retinal implant attempts to provide meaningful vision to blind patients by electrical stimulation of retina. The neural responses of retinal ganglion cell (RGC) are strongly dependent on stimulation parameters such as pulse rate and intensity. Thus, it is necessary to develop a pulse generation strategy so that visual information is transferred reliably to the brain. As an effort toward pulse generation strategy development, we investigated whether the information on visual input can be reconstructed from RGC responses to temporally-patterned stimulation.
Methods: :
Retinas of rd1 mice, as an animal model of degenerated-retina, and normal rabbits, were attached on a 64 channel microelectrode array (MEA). Stimulation pulses were applied to one channel at the center. The amplitudes of pulse trains were modulated to have linearly-increasing, triangular, sawtooth and bandlimited Gaussian random waveforms. The pulse amplitudes were modulated within 20-60 µA range and the pulse rate was varied from 1-10 Hz. Correlation coefficients between pulse amplitude variation and firing rates of evoked RGC spikes were calculated to quantify the accuracy of encoding the input amplitude variation. The input amplitude variation was also reconstructed from multiple single unit RGC spike trains by linear and nonlinear decoding algorithms.
Results: :
Firing patterns of RGCs of normal retinas evoked by stimulation pulses were well modulated by pulse amplitude variation. The decoding of RGC spike trains showed that input intensity variation was encoded in RGC responses faithfully. From the result, optimal parameters for stimulation (amplitude modulation range and pulse rate) were determined. In degenerated-retina, rhythmic firing (8~10 Hz) was observed from spontaneous activities, and the phase of the rhythm was reset by onset of stimulation pulse. Responses to electrical pulse were evoked in addition to the spontaneous rhythmic firing. The evoked components of the RGC activities could also be modulated reliably according to the pulse amplitude modulation.
Conclusions: :
The present results suggest that quantitative analysis and spike train decoding are useful for development of stimulation strategy of retinal implant. Although the RGC response characteristics are drastically changed in degenerated retina, it is expected that the RGC firing can be reliably modulated by temporally-patterned electrical pulse trains.
Keywords: retina • ganglion cells • electrophysiology: non-clinical