Abstract
Purpose :
To measure the reading speed of short sentences at varying contrast levels, with and without spatial white noise. To determine the effect of sinusoidal text motion on the reading speed of this text.
Methods :
Nine normal subjects aged 42±14 read aloud single-line sentences from the IURead database while simultaneously being imaged using the Digital Light Ophthalmoscope – Low Vision (DLO-LV) system.
The DLO-LV combines a custom near-infrared retinal camera with a built-in 5.1” LCD monitor that provides visual stimuli. Imaging is performed at 50 fps over a 28 deg field of view using an 860nm VCSEL source with a time-averaged power of 240uW. The 1920x1080 pixel LCD monitor provides a 25x18 deg visual field of view with a 1 arc min resolution and is controlled as an external PC monitor. Audio was recorded at 8.192 kHz by the DLO-LV.
The visual test was developed using Psychtoolbox-3 in MATLAB (MathWorks, Inc.). A total of 60 unique sentences from the IURead database, with an average of 60 characters and 12 words, were presented to each subject using 20/60 Helvetica font. Three conditions were tested with 5 sentences per text contrast level: no noise or motion, with noise added, and with noise and motion added.
Darker text appeared on a 96 cd/m2 background at contrast levels of 0.12, 0.07, 0.05 and 0.04. Random white noise with a contrast of 0.06 and 4x4 arc min size was added to the background at a 20 Hz refresh rate. Sinusoidal text motion was added with a 50 arc min amplitude and period of 5 sec. Oral reading speed and errors were calculated from the duration of recorded speech.
Results :
The maximum reading speed varied across subjects (104±29 wpm). When 80% of a noise-free sentence was readable, the mean reading speed at each contrast level was 90±30, 58±38, 25±27, 18±23 wpm. Adding noise did not statistically vary the reading speed (p=0.3). Text motion decreased the mean reading speed (p<0.005) by 6±17%, 10±24%, 4±9%, 12±14%. Words near or below the contrast threshold were more easily recognized by subjects at specific visual angles.
Conclusions :
The DLO-LV permits a variety of custom vision research tests using PsychToolbox with simultaneous monitoring of the position of the stimuli with respect to the retina. Reading speeds and contrast threshold varied widely across subjects. Noise did not degrade reading speed, possibly due to a lack of advance practice. Sinusoidal motion of the text lowered the mean reading speed.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.