Stochastic resonance improves vision in the severely impaired

Area of Science:

• Neuroscience
• Vision Science
• Biophysics

Background:

• Sub-threshold visual information is difficult to detect for individuals with impaired vision.
• Stochastic resonance (SR) has previously shown to enhance signal detection in normally sighted individuals.

Purpose of the Study:

• To investigate if stochastic resonance (SR) improves visual information detection in subjects with retinal disorders and impaired vision.
• To determine the optimal noise levels for SR in this population.

Main Methods:

• Six levels of dynamic, zero-mean Gaussian noise were applied to images with varying contrast levels (13 levels).
• Alphabet characters were presented below the subjective visual threshold to 14 visually impaired participants.
• The fraction of recognized letters was measured at baseline and with different noise levels.

Main Results:

• Baseline letter recognition was low (0-0.3) for all subjects.
• Optimal noise levels (sigma: 6-30 grey scale values) significantly increased recognition (peak 0.2-0.8).
• Excessive noise (sigma > 30 GSV) decreased recognition in all subjects.

Conclusions:

• The study successfully replicated the visual processing facilitation of SR paradigms in visually impaired individuals.
• Optimal noise addition via SR enhances visual information processing for those with impaired vision.
• This low-level image manipulation technique shows potential for improving current vision aids and developing future visual prostheses.