Ultrafast Localization of the Optic Disc Using Dimensionality Reduction of the Search Space

Abstract

Abstract. Optic Disc (OD) localization is an important pre-processing step that significantly simplifies subsequent segmentation of the OD and other retinal structures. Current OD localization techniques suffer from impractically-high computation times (few minutes/image). In this work, we present an ultrafast technique that requires less than a second to localize the OD. The technique is based on reducing the dimensionality of the search space by projecting the 2D image feature space onto two orthogonal (x-and y-) axes. This results in two 1D signals that can be used to determine the x-and y-coordinates of the OD. Image features such as retinal vessels orientation and the OD brightness and shape are used in the current method. Four publicly-available databases, including STARE and DRIVE, were used to evaluate the proposed technique. The OD was successfully located in 330 images out of 340 images (97%) with an average computation time of 0.65 seconds.

Automatic optic disc detection using retinal background and retinal blood vessels

“…We incorporate the directional blood vessel information based on the observation that the retinal blood vessels are mostly oriented vertically when they exit the optic disc [12]. Instead of using the L1-norm image gradient [12] that is sensitive to the image degradation, we capture the retinal blood vessel information by an image contrast as follows: Fig.…”

Time Series Case Based Reasoning for Image Categorisation

“…It is technically possible to remove the pixels representing the optic disc in the same way that blood vessel pixels were removed. To achieve this, a variation of optic disc detection using the horizontal and vertical axis of retinal images as proposed in [23] was applied. The retinal blood vessels binary image and the enhanced green channel image was utilised to generate both the horizontal and vertical signals, instead of using the original green channel image [23].…”

Retinal image classification using a histogram based approach

“…To obtain a retinal vessels binary image (Figure 4(d)), the response image is then thresholded using the Otsu's threshold algorithm [21]. The images with enhancement and blood vessels removed are then used to create CB1 An approach where by the 2D retinal image is projected onto two 1D signals (representing the horizontal and vertical axis of the retinal image), similar to that proposed in [22] and [5], was adopted to identify the location of the OD. The number of horizontal and vertical edges together with the intensity value of a retinal image were used as features to localised the OD.…”

“…The number of horizontal and vertical edges together with the intensity value of a retinal image were used as features to localised the OD. Instead of using the original green channel image, as proposed in [22], the authors propose the uses of the retinal vessels image and the enhanced green channel image to generate the (vertical and horizontal) 1D signals. To identify the horizontal location of the OD, a sliding window of double the thickness of the identified main retinal vessel, , and the image height was used to scan the edge maps image from left to right and project the image features of each window onto the horizontal axis 1D signal.…”