Purchase this article with an account.
Carlos AGUILAR, Eric CASTET; Use Of A Gaze-Contingent Augmented-Vision Aid To Improve Reading With Central Feild Loss. Invest. Ophthalmol. Vis. Sci. 2012;53(14):4390.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We investigate whether and how a visual aid based on gaze-contingent augmented-vision can improve visual exploration, especially reading text, when macular vision is lost. Principles underlying our new system were inspired by key concepts in clinical and neuroscience studies of low vision (notably Preferred Retinal Locus, dynamics of spatio-temporal attentional deployment, low-level limiting factors of reading).
A novel real-time gaze-contingent display was developped to allow: a/ simulation of an absolute macular scotoma (10deg x 10deg) with normally-sighted observers, and b/ augmented visual stimulation at eccentric locations outside the scotoma. Gaze location was processed online with an SR-Research eye tracker (500 Hz).The first feature of the augmented vision algorithm consists in shifting the words hidden by the scotoma to a peripheral area adjacent to the scotoma. The second feature is the magnification of the displaced region. Finally, within the displaced region, an "anti-crowding" algorithm adjusts the spaces between letters as a function of crowding extent at each letter location.This algorithm is triggered when the subject pushes a button (the shifted and enhanced region disappears after releasing the button). Monocular oral reading speed was assessed with and without augmented eccentric vision over several one-hour sessions to investigate the effects of learning. Observers had to read French text whose x-height was close to Critical Print Size (1.1°).
Data on the effects of this gaze-contingent system shows a significant improvement of reading speed with the augmented vision algorithm.
We have developed a gaze-contingent augmented-vision system that allows us to design a large variety of novel visual aids and rehabilitation techniques. Current research aims at identifying the most optimal features for our system.
This PDF is available to Subscribers Only