Abstract
Purpose:
In this study we analyzed the visual search pattern of naive subjects during a novel word recognition task. Varied levels of contrast were presented by adding visual noise to word and background in the Laser Scanning Digital Camera Stimulator (LSDC-S). The stimuli were designed to simulate widespread internal visual noise found in ocular and neurological disease, as opposed to focal pathology.
Methods:
The 11 subjects included 4 male and 7 female, ages 22 - 50 yr with visual acuity of 20/20. The stimulus was a 20/200 optotype, 2 - 3 syllable word in Arial font presented centrally in the LSDC-S. A 16 word list was generated from the Low Vision Training Manual (Quillman). The LSDC-S is a slit scanning imager that provides a high contrast, 11 Hz and 36 deg retinal image, 860 nm at 2 mW at the cornea. The visual stimuli were projected by a digital light projector in Maxwellian view, with linked stimulus and retinal image focus. The background illuminance and the contrast for stimuli and background matched Goldmann perimetry, 1.5 log units. Levels of noise, at a fixed spatial frequency, were presented with the words, starting with 9% and then lowered by 3% over 4 levels until the word was recognized. The trial was stopped if recognition did not occur in 6 sec, and the time to recognition was recorded for the final trial. Fixation and sample scan patterns were evaluated from the x, y coordinates of a retinal image feature over time, plus the probability of accuracy.
Results:
Despite the large letter size, this task was difficult for all subjects, who recognized no words with noise > 6% contrast. The average jump in scanning on the x axis was 2.62X larger than for y (p<0.0001). However, the jump size varied, 3.31X to 2.10X for the highest vs. lowest contrasts. One prominent search pattern was to stay at a specific point on the x axis, scan along the y axis, then jump to the next x axis locus. In a subset of the data, the eye movements for two hard words differed from two easy words, (p= 0.006). Smaller jumps were typical for difficult words, “learned” than easier words, “morning.”
Conclusions:
The LSDC-S word recognition task is a novel visual assessment method with varied noise contrast levels. Analysis of fixation reveals both word difficulty and directional differences in search patterns. These data may assist in reading assessment and associated rehabilitation.
Keywords: 760 visual search •
584 low vision •
672 reading