Multiloss Function Based Deep Convolutional Neural Network for Segmentation of Retinal Vasculature into Arterioles and Venules

Figure 5

Vessel thickness inconstancy improvement in the manually prepared label: (a) original image; (b) local contrast normalization; (c) vessel segmentation; (d) manual labelling; (e) visualizing the thickness inconsistency between (c) and (d); (f) results of adjusting results of (c) on (d); (g) optimized labels image after applying enhancement procedure in Figure 4.

(a)

(b)

(c)

(d)

(e)

(f)

(g)