Artikel
MEA-based classification of retinal ganglion cells for bionic vision
Suche in Medline nach
Autoren
Veröffentlicht: | 10. Dezember 2019 |
---|
Gliederung
Text
Objective: Despite significant progress designing retinal implants, how to improve the quality of restored vision using selective activation of different retinal pathways remains unknown. We are exploring whether different Retinal Ganglion Cell (RGC) types can be selectively activated using specific electrical input filters. First, we classify different Ganglion Cell types using a diverse set of visual stimuli. Then, we deliver Gaussian-distributed voltage pulses into the tissue and estimate the electrical input filter of each cell using Spike Triggered Averaging (STA).
Materials and methods: A 60-channel Microelectrode Array (MEA) recorded from mouse RGCs. Visual stimuli were adapted from Baden et al. (Nature 2016): moving bars, contrast and temporal frequency chirps, blue-green flashes, and space-time white noise. Spikes were converted into pseudo-calcium traces for classification according to the Baden cluster with highest correlation. The STA was derived from responses to a 25 Hz train of 1 ms voltage pulses drawn from a Gaussian: mean -800 mV, sigma 280 mV.
Results: We recorded data from 764 RGCs of 5 wild type mice. Data mapped into ~75% of the Baden et al. RGC types. As previously observed, ON RGCs had upward eSTAs and OFF RGCs had downward. Within cell type, eSTAs did not always match.
Discussion: Electrical STAs vary with RGC type, but more precise cluster definitions are needed. For better classification, new definitions should be derived from spike trains.
Acknowledgment: BMBF: 031A308, 01GQ1002; Tistou and Charlotte Kerstan Fdn; DFG: EXC 307.