Automatic Segmentation and Measurement of Vasculature in Retinal Fundus Images Using Probabilistic Formulation

Table 2

Evaluation results on normal and abnormal retinal images from STARE database.

Method	Sensitivity	Specificity	Accuracy
Normal cases
The 2nd observer	0.9646	0.9236	0.9283
Hoover et al. [6]	0.6766	0.9662	0.9324
Soares et al. [13]	0.7554	0.9812	0.9542
Mendonça and Campilho [1]	0.7258	0.9791	0.9492
Proposed	0.7463	0.9664	0.9405
Abnormal cases
The 2nd observer	0.8252	0.9544	0.9425
Hoover et al.[6]	0.6736	0.9472	0.9211
Soares et al. [13]	0.6869	0.9682	0.9416
Mendonça and Campilho [1]	0.6733	0.9669	0.9388
Proposed	0.7034	0.9668	0.9420