Retinal blood vessel extraction using curvelet transform and conditional fuzzy entropy

This work employs multiple thresholds on matched filter response for automatic extraction of blood vessels, specially from a low contrast and non-uniformly illuminated background of retina. Curvelet transform is used first to enhance the finest details along the vessels followed by matched filtering to intensify the blood vessels' response. The conditional fuzzy entropy is then maximized to find the set of optimal thresholds to extract different types of vessel silhouettes from the background. Differential Evolution algorithm is used to specify the optimal combination of the fuzzy parameters. Thresholds thus obtained extract the thin, the medium and the thick vessels from the enhanced image which are then logically OR-ed to obtain the entire vascular tree. Performance evaluated on publicly available DRIVE database is compared with the existing blood vessel extraction methods. Experimental runs demonstrate that the proposed method outperforms the existing methods in detecting various types of vessels.