Optic disk feature extraction via modified deformable model technique for glaucoma analysis

Hybrid Retinal Image Registration Using Mutual Information and Salient Features

“…Typically, these methods can be classified into intensity-based, featurebased, or hybrid-based methods. Feature-based methods extract features from a retinal image first, such as vascular bifurcation points [3], whole vasculature [4], and optic disk [5]. Then, the registration process that finds the best transform parameters is performed by maximizing a similarity measure based on correspondences of the extracted features.…”

Optic disc segmentation using the sliding band filter

“…The outliers of this set are excluded based on two criteria: 1) the relation of its value with the highest value of the filter response; 2) distance to the centroid of the K point set. The criteria for excluding outliers are defined by (9) and (10).…”

“…There are several works on the automatic segmentation of OD in retinal images which can mainly be grouped into four categories, namely template-based methods [4,5,6,7], deformable model methods [8,9,10,11,12,13], morphological-based approaches [14,15,16], and pixel classification methods [17,18]. Within the first category, Aquino et al [4] follow a voting-type algorithm to locate a pixel within the OD as initial information to define a starting sub-image.…”

“…Afterwards, the OD is segmented by means of a deformable contour based on a global elliptical model and on local deformation. In the snake model proposed by Xu et al [9], after each snake deformation, an unsupervised approach labels the contour points as edge-or uncertain-points. Then the classification result is used to refine the OD boundary before repeating the contour deformation.…”

Enhanced retinal image registration accuracy using expectation maximisation and variable bin-sized mutual information

“…While feature-based approaches primarily use optical disk [9], fovea [10] and vascular structural details [11], [12] from the RI, intensity-based techniques focus on pixel intensity information using similarity measures such as cross correlation, phase correlation or MI [6], [13]. MI establishes a statistical relationship between the intensity values of the RI and while it is popular in the medical image registration domain, it is not very effective for RIR because of the aforementioned RI characteristics…”