
Blind people could soon be able to read street signs using an implant that translates the alphabet into Braille and beams an image of the Braille directly to visual neurons at the back of the eye.
The implant is a modified version of a class of devices called retinal prostheses, which are used to restore partial sight to people with retinitis pigmentosa. A degenerative eye disease that kills the photoreceptor cells in the retina, RP tends to affect people in early adulthood and can lead to blindness, but leaves intact the neurons that carry visual signals to the brain.
Prostheses such as the Argus II, manufactured by in Sylmar, California, convert video from a camera mounted on a pair of glasses into electronic signals 鈥渄isplayed鈥 on a 10-by-6 grid of electrodes implanted over a person鈥檚 retina. This gives users a pixellated view of the world, allowing them to distinguish light and dark regions and even detect features such as doorways.
Advertisement
But deciphering letters and words with the prosthesis is slow because of its low resolution. To make this more practical, Thomas Lauritzen of Second Sight and colleagues have come up with a modified version of the Argus II that presents the user with Braille. Since Braille represents letters and numbers as dots in a 3-by-2 grid, it can be displayed using the electrode array of existing Argus implants.
The modified implant was tried out on a Braille-reading volunteer who already uses the Argus II. Tested on single letters and words of up to four letters, transmitted in Braille to the retinal implant, he correctly identified the letters 89 per cent of the time and words 60 to 80 per cent of the time. Longer words should actually be easier to read, Lauritzen predicts, because getting an individual letter wrong creates less confusion than when the word is short.
The user was able to read at a rate of at least one letter per second. By contrast, the pixellated letters of the conventional version of Argus can take wearers tens of seconds to decipher, so whole words can take minutes.
No Braille substitute
The modified system is not intended to replace standard Braille texts: a typical Braille user can read 800 letters per minute by touch. Where the system comes into its own is in situations when no Braille version of a text is available. It could be most useful for reading text in public places, for example, notices and street signs. There are approximately 65,000 people in the US and Europe with severe enough RP to benefit from the prosthesis, says Brian Mech, Second Sight鈥檚 vice-president of business development.
Once the system has been properly tested, the team intends to provide the Braille functionality as a separate mode in the Argus II. In Braille mode, the device would bypass the video processing unit and instead use text-recognition software to identify signs and convert them on the fly into images of Braille. Although it wasn鈥檛 used in the recent study, software exists that can find and read about 90 per cent of signs, Lauritzen says. 鈥淚t鈥檚 already good enough, and it will undoubtedly improve with time.鈥
鈥淪econd Sight have done amazing work for years,鈥 says at Newcastle University in the UK. But the test of any prosthesis is whether it restores abilities to the user. The problem with today鈥檚 visual prostheses is their low resolution, he says.
Sound feedback?
Packing more electrodes into the same space is not currently possible because electrolytic effects make them degrade if they are too close together. 鈥淥ver time the electrodes will fall apart,鈥 says Degenaar. Making the most of the low resolution and using the grid to display Braille is a good idea, he says, but other options should also be explored. If text-recognition software is already so good, then 鈥渨hy not use that to provide auditory feedback rather than Braille?鈥, he asks.
鈥淎nything that potentially leads to new ways to realise vision is very welcome,鈥 says Pete Osborne, chief Braille officer at the in London. Visual prosthetics in general are not trying to replicate sight, he says, and the challenge is to find the best alternative. Besides, Braille was developed as a means of reading by touch. 鈥淲ill it translate to a visual medium? The proof will be in the pudding.鈥
Journal reference: , DOI: 10.3389/fnins.2012.00168