Figure 1.
Overview of the steps in proposed method.
Figure 2.
Image example: a) Two field fundus image after mosaicing b) Green channel image c) Hue channel image.
Figure 3.
Image example: a) Vessel probability image b) Binary image.
Figure 4.
Image example: a) Vessel network b) Vessel tree [Vessel width is enlarged for visualization].
Figure 5.
Graph based description: a) Vessel segment map [Width is enlarged for visualization] b) Representative graph structure.
Table 1.
Algorithm 1: Structural mapping using Dijkstra's graph search.
Figure 6.
Image example: a) Vessel segment map showing the true vessel path b) True vessel path with branches.
Figure 7.
Image example: a) Vessel probability image b) Structural mapping of vessel network.
Figure 8.
Image example: a) Vessel segment map b) Vessel trees with AV crossing c) Vessel trees without AV crossing.
Figure 9.
Fuzzy C-means clustering: a) Cluster formation b) Comparison of mean green channel intensity.
Figure 10.
Image example: a) Structural mapping b) Artery-Venous Classification.
Figure 11.
Metrics: a) Proportion of mis-classified vessel segments, b) Percentage mis-classification per image.
Table 2.
Proportion of mis-classified vessel segments.
Figure 12.
Image example: a) Fundus image b) Vessel probability image c) Structural mapping d) AV Classification.
Figure 13.
Image example: a) Fundus image b) Structural mapping c) Manual AV labeling d) Automated AV Classification.