Robust retinal blood vessel segmentation using line detectors with multiple masks

Robust retinal optic disc and optic cup segmentation via stationary wavelet transform and maximum vessel pixel sum

“…The performance metrics like overlap score ( S ), specificity (SPE), sensitivity (SEN), accuracy (ACC), precision (PRE), F ‐score (), false discovery rate (FDR), area under curve (AUC), Matthew's correlation coefficients (MCC), success(SUC) are calculated as per [10, 37, 38].…”

BAT algorithm inspired retinal blood vessel segmentation

“…In this approach, vessel location map affords better visual quality and a separate process is applied to remove noise and to enhance the segmented vessels. Linear combinations of various scales of line detectors with multiple window sizes of different sizes are intended to extract retinal vessel patterns [29], and this unsupervised method also demonstrates the effectiveness of the proposed method in the presence of false vessel detection regions. The region growing technique and a region-based active contour model [30] with level set implementation is proposed to extract retinal vessels and their results are combined to achieve the final segmentation.…”

Robust retinal blood vessel segmentation using hybrid active contour model

“…Due to less number of freely available fundus images with its gold standard image, unsupervised segmentation methods are profusely used. Different unsupervised segmentation methodology had already been implemented on the retinal image to extract vessels from its background [6–8]. Biswal et al [8] have used a line detector with multiple masks to segment the blood vessels from its foreground.…”

“…Different unsupervised segmentation methodology had already been implemented on the retinal image to extract vessels from its background [6–8]. Biswal et al [8] have used a line detector with multiple masks to segment the blood vessels from its foreground. Roychowdhury et al [23] have used adaptive threshold technique of pixel to extract the vessels.…”

“…Many segmentation methods [8–15] have been proposed in the literature with a substantial contribution to the overall precision of the segmented output. Existing methods have dealt with some of the problems such as false vessel detection, overlapping of two crossing blood vessels, missing of tiny blood vessels and so on.…”