Fig 1.
Retinal surface with three major structures.
Table 1.
Description of participants.
Fig 2.
Example display from the software tool developed for the experiment.
Fig 3.
Instructions given to participants.
Fig 4.
Total track length for sample images.
(a)-(b) Gaze-tracks for two different subjects for an image with a lesion (marked with green circles) in the periphery. One subject takes longer route than another. (b)-(c) Gaze-track for an image with peripheral and more centrally located lesion. Image with a lesion at periphery(center) requires longer(shorter) total track length.
Fig 5.
Dwell duration is proportional to the radius of the disc representing fixation. Lesion locations are marked with green ovals. (a)-(b) Gaze-track for two different subjects for an image. Very different dwell pattern shows total dwell duration depends on subjective behavior. (b)-(c) Gaze-track for two different stimuli images. Gaze-track for image with subtle(prominent) lesion has more(less) total dwell duration.
Fig 6.
CS values for 2240 responses.
Table 2.
Average coefficient of scanning for various image category and response pair.
Fig 7.
Accuracy achieved with dwelling strategy.
Fig 8.
Accuracy achieved with tracing strategy.
Fig 9.
Correlation between track length and revisits.
Fig 10.
Retinal zones recommended by ETDRS.
Fig 11.
Average transition matrices for different expertise-groups.
Top/bottom row: abnormal/normal cases. Left to right: consultants, fellows, residents/optometrists, novices. Each transition matrix is scaled and color-coded separately.
Fig 12.
Average dwell maps for different expertise-groups.
Top/bottom row: abnormal/normal cases. Left to right: consultants, fellows, residents/optometrists, novices. Each dwell map is scaled and color-coded separately.
Fig 13.
Transition matrix (a) for normal cases (b) for abnormal cases (c) average transition matrix TMavg (d) net transition matrix TMnet (e) binary neighborhood matrix NM (e) conditioned net transition matrix TMc.
Fig 14.
Gaze-pattern of one subject for 4 images.
(a), (b) visual search for abnormal cases is terminative (c), (d) visual search for normal cases is exhaustive.
Fig 15.
(a) Optimal transition pattern (b) Optimal dwell map (c) Optimal scanning strategy.
Transitions in optimal scanning strategy are guided by optimal transition pattern and dwelling by dwell map. The radius of circle represents the amount of required dwelling.
Fig 16.
(a) Average accuracy and (b) response time for scanning strategies which are at equal levenshtein distance from optimal strategy.