Abstract
The scanning laser ophthalmoscope (SLO) is an instrument of great value for observing viewing behaviour and characteristics of fixation in macula disorders; it also may have potential in assessing visual rehabilitation in affected individuals. It is important to understand the viewing strategies used by individuals with a central scotoma and eccentric fixation and to determine whether these strategies have changed as a result of any eccentric viewing training. We describe a method based on searching for multiple sectors in relative correspondence in consecutive SLO fundus images. In this method, successive fixation positions, and consequently viewing strategy, during the reading of isolated words and text passages can be extracted from SLO images. We found the maximum cumulative error in determining fixation position over 1000 consecutive frames to be 5 pixels (approximately 15 min arc) using fundus images from an individual with a macular lesion and clearly visible retinal features. The reconstruction of fixation sequence was found to closely resemble real time or frame-by-frame viewing of the videotaped fundus images from a series of reading tasks in over 50 patients with central scotomas. This method can also be used to assess reading behavior through an SLO in ocularly-healthy individuals.