Purchase this article with an account.
Nairouz Farah, Chen Abraham, Liron Gerbi, Zeev Zalevsky, Yossi Mandel; Active sensing and reading performance in simulated prosthetic vision. Invest. Ophthalmol. Vis. Sci. 2018;59(9):4565. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To study whether active sensing can induce visual performance enhancement of simulated prosthetic vision similar to the well-known effect of saccadic eye movements.
To simulate prosthetic vision and to study reading performance, english words randomly drawn from a database were converted to phosphenized words at different phosphene densities (0.75-2 cycles per degree (CPD)) and contrasts (100%,50%,25%). Three sensing paradigms were implemented: "active sensing" in which the subject actively scanned the presented words using the mouse; passive scanning produced by computer controlled horizontal movements of words; and “no scan” paradigm. In addition, we studied the effect of enabling zoom controlled by the subject. We compared the recognition rates and reading speed for the various test paradigms and analyzed the scanning path employed by the subjects.
An increase in recognition rate with increasing contrast and phosphene density was observed, reaching a plateau at phosphene densities higher than 1.25 CPDs. Reading accuracy increased by a factor of up to 1.3 with active scanning as compared to no or passive scanning for the range of 0.8-1 CPDs (p<0.01). Moreover, we observed that zooming significantly enhanced word recognition rate by up to 14 fold. Interestingly, for the same set of investigations reading speed was not affected. Scanning path analysis revealed a preference to horizontal directions with scanning amplitude, speed and time invested in rapid scanning movements decreasing with phosphene density, indicating that active scanning is in use mostly at lower phosphene densities.
Active sensing can enhance prosthetic vision reading performance. Moreover, our results suggest that image processing algorithms, for the correction of low contrast or small letter text, can be very beneficial in enhancing prosthetic vision reading performance. More work is needed at investigating the applicability of these findings , and for estimating the effect of learning and training on prosthetic vision reading.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.
This PDF is available to Subscribers Only