Biomedical Engineering Reference
In-Depth Information
only when considered from a low-resolution perspective. Recent work by Warren and
Fernandez (2001) has shown that it is possible to use the relative location of a neuron in
the cortex to predict where spots of light projected onto the visual space will excite that
neuron. However, this can be done with an accuracy of only about 0.5
◦
of visual angle.
Signal processing must accomplish a number of tasks. First, the electronics must
transform the video signal into a set of discrete signals—one for each electrode. Next, the
processor must be able to generate the correct amplitude and dynamic range of stimulus
levels for the prosthesis, irrespective of the ambient light level. This is an automatic gain
function that emulates the adaptive properties of both the photoreceptors and the pupil.
Part of this process involves compression of the dynamic range.
The final function of the signal processor is to map the spatial output of the image
to the electrode array so that vertical lines are perceived as vertical. The degree of this
remapping depends on the degree of apparent randomness in the visuotopic organization
at the implant site.
7.9.3.3 Telemetry and Power Interface
A visual prosthesis must receive information about the visual scene as well as power and
data to run the electronics and subsequently stimulate the retina or cortex to generate the
corresponding visual image. Given that it is undesirable for the purposes of long-term
implantation to have wires penetrating the body or embedded batteries that have a limited
life, it is necessary to send the visual signal and power to the implant wirelessly. It is only
with recent advances in microelectronic technology that this has become feasible. If the
prosthesis is retinal, then the carrier signal can be either light or radio frequency (RF).
However, if it is embedded in the optic nerve or the visual cortex, then RF is generally
used. It is envisaged that similar principles to those used for cochlear implants will be
used as they have been successful in experimental applications.
The processes involved to accommodate both power and bidirectional telemetry have
been discussed in Chapter 5. If the internal prosthesis is retinal, it would be convenient to
mount one of the coils within the frame of a pair of eyeglasses and the other coil within
the eye. This will ensure good alignment, and small variation of range, with a resultant
high-efficiency signal and energy transfer. Radiated power will have to be limited to
minimize local heating, and therefore the embedded parts of the prosthesis will have to
be extremely efficient. For cortical implants there is no convenient means to mount the
external prosthesis really close to the implant. However, behind-the-ear mounting as is
used in cochlear implants would still limit the length of the internal wiring and provide a
reasonable compromise.
Other issues include the required high reliability, and in the future, as the array sizes
of the prosthetics increases, the telemetry will have to support high-bandwidth commu-
nications. This will be mitigated to some extent by improvements in image compression
along with their associated encoding and decoding techniques.
7.9.3.4 Neural Stimulator
Stimulators can be based on existing designs that have been developed for other applica-
tions such as those for cochlear implants. These are typically external and have limited
bandwidth due to the small number of electrodes involved, so their performance would
have to be enhanced significantly to accommodate the requirements of a visual prosthesis.
Current thinking (Finn and LoPresti, 2003) is that stimulators will be digital to facilitate
upgrades and to minimize power dissipation. They will include sufficient memory to
