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

Table 2

Optimized multiloss function judgment.

ROI	Total# of Pixels	Label class	Total# of Red Pixels	Total# of Blue Pixels	Artery Probability	Vein Probability	Maximum Probability	Multi- Loss Judgment
Seg1	200	Artery	120	80	0.600	0.400	Red	Artery
Seg2	500	Vein	400	100	0.800	0.200	Blue	Vein
Seg3	700	Vein	100	600	0.100	0.900	Blue	Vein
Seg4	457	Vein	46	411	0.100	0.900	Blue	Vein
Seg5	458	Artery	320	137	0.700	0.300	Red	Artery
Seg6	459	Vein	0	459	0.000	1.000	Blue	Vein
Seg7	460	Vein	184	276	0.400	0.600	Blue	Vein
Seg8	320	Vein	43	277	0.134	0.867	Blue	Vein
Seg9	230	Artery	146	84	0.634	0.367	Red	Artery
Seg10	543	Artery	326	217	0.600	0.400	Red	Artery
Seg11	544	Artery	508	36	0.934	0.067	Red	Artery
Seg12	545	Artery	529	18	0.967	0.034	Red	Artery