Definition of the problem: Our goal is to determine the minimum of information necessary for elementary reading, using a retinal implant. This concerns particularly the fragmentation (pixellisation) of the presented image and its position in the visual field. Fragmentation corresponds to the number of electrodes available, the position of the image in the visual field is equivalent to the site of the implant on the retina.
Material and methods: 10 degrees x 10 degrees windows, containing isolated words or letters, were presented to six healthy subjects on a computer screen. A coupling between the computer and an eye tracker stabilizes these images in an area of the visual field. This coupling constantly corrects the position of the image on the screen according to the direction of gaze.
Results: 1) A rapid decrease of the performance is observed at a certain threshold of pixellisation, dependent on the eccentricity of presentation of the images. 2) In central vision, about 400 pixels are sufficient to recognize 80% of the four-letters words. At 10 degrees of eccentricity, about 1225 pixels are needed. 3) An acceptable comprehension of a text (identification of four words out of five), is impossible at eccentricities higher than 10 degrees 4) About 50 pixels are sufficient for a satisfactory recognition of isolated letters, independently of their eccentricity.
Conclusion: These data validate the method of investigation and provide valuable indications regarding minimal visual requirements in prosthetic vision.