Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Sep 1;118(3):1457-1471.
doi: 10.1152/jn.00750.2016. Epub 2017 May 31.

Activation of ganglion cells and axon bundles using epiretinal electrical stimulation

Lauren E Grosberg  1 , Karthik Ganesan  2 , Georges A Goetz  3 , Sasidhar S Madugula  3 , Nandita Bhaskhar  2 , Victoria Fan  3 , Peter Li  4 , Pawel Hottowy  5 , Wladyslaw Dabrowski  5 , Alexander Sher  6 , Alan M Litke  6 , Subhasish Mitra  2 , E J Chichilnisky  3
Affiliations

Activation of ganglion cells and axon bundles using epiretinal electrical stimulation

Lauren E Grosberg et al. J Neurophysiol. .

Abstract

Epiretinal prostheses for treating blindness activate axon bundles, causing large, arc-shaped visual percepts that limit the quality of artificial vision. Improving the function of epiretinal prostheses therefore requires understanding and avoiding axon bundle activation. This study introduces a method to detect axon bundle activation on the basis of its electrical signature and uses the method to test whether epiretinal stimulation can directly elicit spikes in individual retinal ganglion cells without activating nearby axon bundles. Combined electrical stimulation and recording from isolated primate retina were performed using a custom multielectrode system (512 electrodes, 10-μm diameter, 60-μm pitch). Axon bundle signals were identified by their bidirectional propagation, speed, and increasing amplitude as a function of stimulation current. The threshold for bundle activation varied across electrodes and retinas, and was in the same range as the threshold for activating retinal ganglion cells near their somas. In the peripheral retina, 45% of electrodes that activated individual ganglion cells (17% of all electrodes) did so without activating bundles. This permitted selective activation of 21% of recorded ganglion cells (7% of expected ganglion cells) over the array. In one recording in the central retina, 75% of electrodes that activated individual ganglion cells (16% of all electrodes) did so without activating bundles. The ability to selectively activate a subset of retinal ganglion cells without axon bundles suggests a possible novel architecture for future epiretinal prostheses.NEW & NOTEWORTHY Large-scale multielectrode recording and stimulation were used to test how selectively retinal ganglion cells can be electrically activated without activating axon bundles. A novel method was developed to identify axon activation on the basis of its unique electrical signature and was used to find that a subset of ganglion cells can be activated at single-cell, single-spike resolution without producing bundle activity in peripheral and central retina. These findings have implications for the development of advanced retinal prostheses.

Keywords: axon bundles; brain-machine interface; raphe; retinal electrophysiology; retinal ganglion cells; retinal prosthesis.

PubMed Disclaimer

Figures

Fig. 1.
Fig. 1.
Semiautomated method for detecting somatic activation. A, top row: mean-subtracted waveforms recorded on the stimulating electrode immediately following electrical stimulation, at four stimulation amplitudes. Bottom row, at the same amplitudes, the coefficients for each trial corresponding to the first 2 principal components of the recorded waveforms form distinct clusters. Estimated cluster centers are indicated by black circles. Red (gray) waveforms and points indicate trials that were identified automatically as containing (not containing) spikes. B: a cumulative Gaussian function was fitted to the probability of activation across trials computed from the data in A. Error bars represent ±1SD of 100 probabilities computed after the trials were resampled with replacement 100 times. The activation threshold (0.23 μA) was defined as the stimulation amplitude that produced 50% activation probability according to the fitted function. C: electrode configuration for the waveform comparison. Recordings from the stimulating electrode (SE) and the surrounding electrodes were inspected. Numbers correspond to the waveform plots in D and E. D: the consistency of the waveform shape at consecutive stimulation amplitudes was verified to ensure that the elicited waveform was produced by a single cell. For each stimulation amplitude (indicated by the shade of red in the plot), the mean of the nonspiking trials was subtracted from the mean of the spiking trials on the stimulating and surrounding electrodes to extract the electrically elicited spike waveform. E: spike templates (electrical images) obtained with visual stimulation revealed 6 RGCs recorded at the stimulating electrode, identified by their unique neuron labels a–f. The template waveform (b) that matched the electrically elicited waveforms shown in D is highlighted in red.
Fig. 2.
Fig. 2.
Bidirectional propagation of electrically evoked responses. A: fluorescence image shows the density and arrangement of RGC axon bundles with respect to the electrode array. Arrow indicates the stimulating electrode in C and D. B: electrical images (EIs) from 3 RGCs in a single retina obtained with visual stimulation (no electrical stimulation). All of the axons run in the same direction toward the optic disk. Waveforms on the electrodes indicated with numbers are associated with the shaded RGC. C and D: unidirectional (C) and bidirectional (D) signal propagation after electrical stimulation at the electrode shown by a red circle. Note similarity of unidirectional (C) image to the EI from the shaded cell shown in B. The amplitude of the waveform shown for electrode 6 in A–D was reduced by a factor of 2 relative to the scale bar.
Fig. 3.
Fig. 3.
Axon recruitment in bidirectional axon bundle signal. A: 2-dimensional interpolation of the maximum amplitudes recorded in the 5 ms following stimulation at the electrode indicated by the circle (mean of 25 trials). Lines indicate cross sections over which amplitudes are shown in B and C. The “downstream” signal direction toward the optic nerve is indicated by the arrow. B: voltage recordings along the cross sections shown in A. Numbers correspond to the profile line. Colors indicate stimulation amplitude for each voltage profile shown. C: the maximum amplitude of the profiles (averaged across trials) shown in B increase progressively with stimulation current. Data from individual trials are shown in gray. The approximate number of axons activated, based on the amplitude of a typical axon in this retina, is shown for comparison (far right).
Fig. 4.
Fig. 4.
Axon bundle activation threshold map exhibits spatial variability. A: in a single recording, bundle threshold for each electrode is indicated by the color of the dot representing that electrode and the inverse of the dot size. B: for 6 preparations from 5 retinas (3 and 4 represent different preparations from the same retina), box-and-whisker plots summarize the medians, quartile values, and ranges of the bundle thresholds for all electrodes in the recording. Data from retina 1 are shown in A.
Fig. 5.
Fig. 5.
Automated and manual axon bundle threshold estimation yield similar results. A: scatter plot compares the bundle activation thresholds identified by the algorithm and manual inspection. Color is proportional to the logarithm of the density of points. The correlation between the two methods was 0.93 for the 1,885 tested electrodes shown (5 preparations). B: histogram shows the ratio of the thresholds identified by the algorithm and the manual observer, revealing that 88% of the estimates obtained from the two approaches were within ±10% (3 central bins). C: scatter plot shows variability in manual inspection. The correlation of threshold values reported by two observers was 0.97. D: histogram shows the ratio of the thresholds identified by the two observers, revealing that 95% of estimated threshold values were within ±10%.
Fig. 6.
Fig. 6.
RGC and bundle activation threshold comparisons. A: colored histograms show RGC activation thresholds for 3 preparations with eccentricities of 48.2°, 58.1°, and 58.1°. Open histograms show axon bundle thresholds for the electrodes used to stimulate the same RGCs. B: scatter plots show the RGC somatic threshold vs. bundle threshold at that electrode. Color is proportional to the number of data points in a local region. In aggregate, 45% of all electrodes that were able to activate a single RGC (i.e., 17% of all electrodes on the array) were able to do so without bundle activation. Data are displayed from 592 electrodes that activated RGCs at their somas from 3 retinas.
Fig. 7.
Fig. 7.
Impact of spatial resolution on selective activation of RGCs. Histograms show the number of electrodes at a given distance to the corresponding estimated RGC activation site for 3 retinal preparations. Color indicates electrodes with somatic activation threshold below bundle threshold; gray indicates the reverse.
Fig. 8.
Fig. 8.
Visual receptive fields of RGCs that can be activated with and without bundle activation. Data in A–C, D–F, and G–I correspond to the retinal preparations described in the text with eccentricities of 48.2°, 58.1°, and 58.1°, respectively. A, D, and G represent the receptive fields of the cells that can be activated at their somas without activating other nearby somas. Receptive fields are separated into ON and OFF parasol cells, ON midget cells, and other cells, which include OFF midget cells, small bistratified cells, and cells for which the anatomical identity is unknown. B, E, and H represent the receptive fields of the cells that can be activated without activating bundles. C, F, and I show zoomed images of axon bundles in each preparation, with respect to a grid of electrodes (green overlay, arbitrary alignment) with spacing equal to that used in the experiments.
Fig. 9.
Fig. 9.
Stimulation in the raphe region. A: immunolabeling reveals the raphe region of the primate retina, with a relatively low density of axons. B and C: histogram and scatter plot, respectively, show the RGC somatic threshold vs. bundle thresholds at the same electrodes (see Fig. 6). Seventy-five percent of all electrodes that were able to activate a single RGC (i.e., 16% of all electrodes on the array) did so without bundle activation.
Fig. A1.
Fig. A1.
Bundle paths and signal growth identified with graph traversal. A: electrodes on the array are represented by circles, colored by the minimum recorded voltage on each electrode during the 5 ms following electrical stimulation. Start and end points were chosen as the electrodes on the array borders with the largest recorded signal, and a path (red) was identified joining these electrodes. B: mean voltage deflections recorded on electrodes in the bundle path (A) as a function of stimulation amplitude were used to identify the bundle activation threshold. Bundle activation was identified when the squared difference of sequential points on the curve exceeded a fixed threshold. Data are from the same preparation as Fig. 3.
Fig. A2.
Fig. A2.
Illustration of graph creation, partitioning, and bidirectional propagation testing to determine bundle activation thresholds. A: mean waveforms recorded on all electrodes after stimulation at each electrode were used to construct a complete, directed, weighted graph (left). The graph is transformed by zeroing all edges not connected by a path produced by electrical stimulation, and then symmetrized (middle). Weights in the resulting undirected graph (right) are represented by the thickness of connecting lines (thicker line corresponds to a stronger connection). B: initial 512 × 512 adjacency matrix for one electrical stimulation data set (left). Entries in the preceding adjacency matrix not connected by a path were zeroed (middle). The adjacency matrix was symmetrized and the graph partitioned on the basis of its modularity (right). Groups of electrodes that were identified are outlined using distinct colors. C: groups of electrodes are revealed using the same colors as B. D: immunohistochemistry image of axon bundles for same preparation as C, with electrodes overlaid. The groups of electrodes from C conform to the geometry and direction of labeled axon bundles. E: visualization of bidirectional propagation testing. Left, the group of one particular stimulating electrode (black) is bisected, resulting in two bands (red and blue). Right, stem plots show displacement and angle of movement of the center of mass of elicited activity in each of the two bands. Consecutive time samples with bidirectional propagation are highlighted in red and blue, respectively.
Fig. A3.
Fig. A3.
Sensitivity of automated bundle detection to individual axon signals in synthetic data. A: when the synthetic single axon signal (see materials and methods) was added to 100% of trials one amplitude step lower than the originally estimated bundle threshold, it was detected correctly in 78% of cases tested. B: when the single axon signal was added to 50% of trials, it was detected correctly in 62% of cases tested.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Abramian M, Lovell NH, Morley JW, Suaning GJ, Dokos S. Activation and inhibition of retinal ganglion cells in response to epiretinal electrical stimulation: a computational modelling study. J Neural Eng 12: 016002, 2015. doi:10.1088/1741-2560/12/1/016002. - DOI - PubMed
    1. Ahuja AK, Dorn JD, Caspi A, McMahon MJ, Dagnelie G, Dacruz L, Stanga P, Humayun MS, Greenberg RJ; Argus II Study Group . Blind subjects implanted with the Argus II retinal prosthesis are able to improve performance in a spatial-motor task. Br J Ophthalmol 95: 539–543, 2011. doi:10.1136/bjo.2010.179622. - DOI - PMC - PubMed
    1. Boinagrov D, Pangratz-Fuehrer S, Goetz G, Palanker D. Selectivity of direct and network-mediated stimulation of the retinal ganglion cells with epi-, sub- and intraretinal electrodes. J Neural Eng 11: 026008, 2014. doi:10.1088/1741-2560/11/2/026008. - DOI - PMC - PubMed
    1. Chan LL, Lee EJ, Humayun MS, Weiland JD. Both electrical stimulation thresholds and SMI-32-immunoreactive retinal ganglion cell density correlate with age in S334ter line 3 rat retina. J Neurophysiol 105: 2687–2697, 2011. doi:10.1152/jn.00619.2010. - DOI - PubMed
    1. Chichilnisky EJ. A simple white noise analysis of neuronal light responses. Network 12: 199–213, 2001. doi:10.1080/713663221. - DOI - PubMed

Publication types

LinkOut - more resources

-