Abstract
Purpose.:
To examine the visual evoked potentials (VEP) and electroretinograms (ERG) generated during electrical stimulation of the human optic nerve using the optic nerve visual prosthesis.
Methods.:
Two volunteers blind from retinitis pigmentosa (RP) and with no light perception each received a chronically implanted optic nerve visual prosthesis. Cortical evoked potentials were recorded using 16 scalp electrodes, and antidromic ERGs were obtained using DTL electrodes while the optic nerve was electrically stimulated. The results were compared with flash and eye surface electrical stimulation results in normal-sighted control subjects.
Results.:
The VEPs obtained in our two volunteers with implants had a waveshape similar to that obtained in normal-sighted volunteers during flash stimulation, but latency was reduced by approximately 25 ms. The VEPs recorded during surface eye stimulation are similar in both normal-sighted and RP volunteers. The VEPs were compared at sub- and supra-threshold stimulation strength and with different electrode configurations. Finally, the antidromic ERG recordings obtained in our implanted volunteers show a unique inner retinal potential generated by retrograde stimulation of the eye from the optic nerve.
Conclusions.:
Evoked potentials can be used to examine how a visual prosthesis generates visual sensations. This provides an objective means to investigate various aspects of the visual prostheses.
A visual prosthesis is a device that aims to restore a meaningful sense of visual perception to blind volunteers by electrically stimulating their visual system. There are many devices being researched that differ according to the position along the visual system used to interface the stimulating electrode. They are broadly divided into the subretinal,
1–3 epiretinal,
4 optic nerve (Sakaguchi H, et al.
IOVS 2008;49:ARVO E-Abstract 4044),
5 and visual cortex prostheses.
6–8 For a review of this topic, see Margalit et al.
9 or Veraart et al.
10 The optic nerve visual prosthesis (ONVP) uses a self-sizing spiral cuff electrode to electrically stimulate the optic nerve and thus create the perception of small flashes of light called phosphenes. By controlling the stimulation parameters, it is possible to vary the position of these phosphenes
11 and thus convey information regarding the surrounding environment.
12 Two volunteers each received a chronically implanted ONVP. In the first, the electrode was implanted on the intracranial section of the optic nerve.
5 In the second, it was implanted intraorbitally.
13
Thus far, the ONVP has been used in psychophysical experiments such as pattern recognition,
14 identifying and localizing objects placed on a table,
15 and mobilizing in foreign environments. Results are encouraging and show that the ONVP can be used for visual rehabilitation.
16 Similar psychophysical results are now being reported from other research teams with chronically implanted human volunteers.
4 However, little work has been published on the evoked potentials generated by a visual prosthesis in humans. All the work on evoked potentials relates to animal studies during the initial trial phases of visual prosthesis development.
This study investigated the evoked potentials generated electrically in volunteers implanted with visual prostheses. A comparison is made between the electrical evoked potentials in visual prosthesis volunteers and flash evoked potentials in normal-sighted control subjects. Thresholds for activating the optic nerve are objectively measured. Furthermore, evoked potentials can be electrically generated in both normal-sighted and blind volunteers by a noninvasive technique previously reported by us using eye surface stimulation.
17 This allows direct comparison to be made when the visual pathway of the two groups are stimulated electrically in the same manner.
Optic Nerve Visual Prosthesis.
For a detailed description of the ONVP, see Delbeke and Veraart
18 or Delbeke et al.
19 In brief, the ONVP is made up of external and implantable components. The main external component is the processing unit, which receives stimulation commands from a computer running a data acquisition software program (Labview, Vernier, Beaverton, OR). The computer and external signal processor were connected wirelessly (Bluetooth). The processing unit is responsible for converting the commands into stimulation pulses, which are then transferred to the implantable neurostimulator by a radiofrequency telemetric link. This link allows data to be transferred across the scalp of the volunteer and is achieved using two antennas that are aligned and connected using magnets. The data are transferred at a rate of 3 Mb/s, and 250 mW power is generated for the implantable components. The neurostimulator, which is surgically embedded into the parietal cranium of the volunteer, works in the same way as the stimulators for cochlear implants. It has individually addressable current sources that send the stimulation pulses to the optic nerve cuff electrode.
To reduce the amount of noise in the recordings, the volunteer was placed inside a Faraday cage. The external components of the prosthesis were placed outside the cage, and a twisted-pair cable was passed from the external processor, through a grating in the cage, to the external antenna of the telemetric link. Preliminary experiments showed that this setup dramatically reduced the level of noise in the recordings.
In each recording session, the optic nerve was stimulated 100 times at 0.3 Hz with single-charge recuperated pulses with a ratio of 1:9. Each of the contacts around the optic nerve was used in turn; the pulse amplitudes were varied from 92 μA to 1040 μA, and the durations were varied from 213 μs to 426 μs.
Surface Stimulation.
Light Stimulation.
The recording method shown in this study was derived from the International Society for Clinical Electrophysiology for Vision standards for recording VEPs and ERGs.
20,21 One of the main difficulties in obtaining good recordings in our volunteers with implants was that the visual prosthesis itself creates electrical noise that could easily drown the signal. By placing the external components of the visual prosthesis outside a Faraday cage and the volunteer inside, the quality of the recordings improved dramatically. It was also necessary to cut the radiofrequency transmission to the implanted neurostimulator after the optic nerve was stimulated to minimize the impact of the stimulation artifact.
All cortical evoked potentials, whether generated using flash stimulation in normal-sighted controls, eye surface stimulation, or optic nerve stimulation in the implanted volunteers, had similar waveshapes. In this study we found that maximal occipital cortex activation took place at around 75 ms after electrical stimulation of the optic nerve. A 2-ms shorter latency was, in fact, found when stimulation was performed with the intracranial rather than the intraorbital electrode because it was further downstream in the visual pathway.
Evoked potentials generated by eye surface stimulation are thought to arise from activation of the inner retina. Potts et al.
22 showed that rats with hereditary outer retinal degenerative conditions have absent ERG and VEPs but near-normal electrically evoked cortical potentials. Similarly, in humans, the electrically evoked response to surface eye stimulation is near normal in patients with rod/cone dystrophy
23 but reduced or absent in patients with inner retinal abnormalities such as central retinal artery occlusion
24 and optic nerve disease.
25 Electrically evoked potentials using surface eyelid stimulation shown in this study demonstrated a striking similarity between an RP volunteer and a normal-sighted subject. In both cases, the inner retina was noninvasively stimulated, and both groups had a peak response at around 80 ms.
It will be interesting to see how these recordings compare with those generated using an epiretinal or a subretinal visual prosthesis. It is likely that an epiretinal device will produce a latency similar to surface eye stimulation because both techniques stimulate the inner retina.
26,27 Similarly, subretinal or suprachoroidal stimulation of the outer retina should produce evoked potentials with longer latencies.
Our second volunteer, who had the intraorbital cuff electrode, had a higher stimulation threshold for perceiving phosphenes because of the position of the intraorbital cuff electrode, which was outside the meninges and not in direct contact with the nerve, as it is intracranially. Nevertheless, phosphenes can be created with both devices, and in both volunteers VEPs could be measured. VEP amplitude, in keeping with the subjective brightness described by our two volunteers, was smaller in the second volunteer than in the first. A greater stimulation artifact was produced in our second volunteer, but the waveshapes thereafter were similar in shape (
Fig. 2). Once thresholds were reached, the amplitude of the VEP increased less in our second volunteer than in our first volunteer when stimulation strength was increased. Subjectively, the first volunteer also described more marked changes in the brightness, location, and size of the perceived phosphenes when increasing the stimulus amplitude.
When stimulating at different locations around the nerve, the waveshape and, hence, the equivalent source dipole changed very little in location or orientation even though, during stimulation with different contacts, both volunteers perceived phosphenes at different locations. All phosphenes elicited with the ONVP occurred within a limited central area of the visual field. In this study, single-pulse stimuli were used that produced phosphenes at the edge of this electrically inducible visual field.
11 Even so, with the setup used for this study, it was not possible to discriminate the different phosphenes elicited from evoked potentials alone.
The VEP recordings in our implanted volunteers were used to objectively estimate the thresholds for stimulating the optic nerve. By varying the amplitude of stimulation, it was possible to establish the threshold for activating the optic nerve. These thresholds form an integral part of the predictive models used to calculate the stimulation parameters required for the generation of phosphenes within discrete areas of the visual field.
11 Until now, these thresholds have been obtained psychophysically, and there was, in this study, good correlation with the objectively measured thresholds. Recently, it has been found that the threshold for stimulating the optic nerve diminishes over time (Delbeke J, manuscript submitted). Thus far, only our first volunteer has been implanted long enough for this phenomenon to be observed.
Our implanted volunteers showed unique ERGs that were generated by retrograde electrical stimulation of the optic nerve. There did not appear to have been any latent secondary activation of the inner retina, as seen in epiretinal devices.
26 However, further animal experiments will be needed whereby the optic nerve can be stimulated and recorded simultaneously to demonstrate this conclusively.
Supported by European Commission CEU Grant IST-2000-25145 (OPTIVIP), FMSR Grant 3.4590.02, and Walloon Region of Belgium Contract 114645, and by an FSR grant from the Université catholique de Louvain (MEB).
Disclosure:
M.E. Brelén, None;
V. Vince, None;
B. Gérard, None;
C. Veraart, Neurotech S.A. Belgium (I, C);
J. Delbeke, Neurotech S.A. Belgium (I, C)