Large-scale Integration–Based Stimulus Electrodes for Retinal Prosthesis - Artificial Sight: Basic Research, Biomedical Engineering, and Clinical Advances - page 158

filtering is that the number of required logic gates is small because there is no
need for adders or multipliers.
We have also implemented a binarization circuit based on an asynchronous
counter with N-bit D-FF. The input of the D-FF of the MSB is fixed to HIGH.
When 2 N − 1 pulses are input, the output of the counter turns HIGH from LOW.
This means that the counter works as a digital comparator with a fixed threshold
of 2 N − 1 .
According to the proposed architecture, we have designed and fabricated
a PFM-based retinal prosthesis device with 16 16 pixels. Figure 8.6 shows

microphotographs of the fabricated chip, peripheral circuits, and its pixel. As
shown in Figure 8.7, each pixel has a PFM photosensor, the above-described
image processing circuits, stimulating circuits, and a stimulus electrode, i.e. this
chip can stimulate retinal cells. We used this chip for in vitro electrophysiological
experiments of stimulating retinal cells, as described in the following section.
Here, we present only the image processing results. Figure 8.5b shows the
experimental results of image processing using the chip: an original image, edge
enhancement, and blurring. These results clearly demonstrate that the proposed
architecture works properly.

×

Figure 8.6. Microphotographs of the fabricated chip, peripheral circuits, and its pixel.