Duration to find the target within the image compared to saccades per second (a, b) and saccade amplitude (c, d) in the real-world search task for the n = 40 controls and n = 40 patients respectively. Univariate linear regression was performed for saccade rate compared to search duration in both the control (a) and patient (b) groups. Further to this, exponential regression (a 3 parameter exponential regression where x is represented by saccade rate) was applied to the latter, which is approximately 8 million times more probable when comparing its AIC to that of the linear model (b) . The red coloured symbols within the patient plots correspond to those patients highlighted in Figure 3. Note, for example, that patients represented by the red circle and triangle symbols have similar levels of VF defect severity (see Figure 3), but very different search duration performance.
Smith et al. BMC Ophthalmology 2012 12:45 doi:10.1186/1471-2415-12-45