An unsupervised approach to improve contrast and segmentation of blood vessels in retinal images using CLAHE, 2D Gabor wavelet, and morphological operations

Abstract

Purpose Retinopathies are the leading cause of eyesight loss, especially among diabetics. Due to the low contrast of blood vessels in fundus images, the visual inspection is a challenging job even for specialists. In this context, this work aims to implement image processing techniques to support contrast enhancement and segmentation of retinal blood vessels. Methods The initial proposal consisted only of green channel separation, contrast limited adaptive histogram equalization, and 2D Gabor wavelet and mathematical morphology. The new proposal includes the edge and mask detection and the vessel enhancement 2D to preserve image's characteristics. The development and validation of this work, in MatLab® environment, involved 40 images from Digital Retinal Images for Vessel Extraction (DRIVE), 20 images from Structured Analysis of the Retina (STARE), and 45 images from High-Resolution Fundus (HRF) database. Results In the unsupervised method context, the proposal presented the best performance regarding sensitivity and second place for balanced-accuracy on all databases. A subjective validation involving eleven ophthalmology professionals showed higher levels of acceptance (above 80%) after contrast limited adaptive histogram equalization (CLAHE) and vessel enhancement 2D steps and 75.5% for overall quality system. Conclusion The main contributions refer to the inclusion of techniques for automatic mask detection, image edge removal, and suppression of vessels background to improve the retinal vessels segmentation process. In addition, this work made a computational interface named "Retinal Lab-A Tool for Fundus Image Analysis" available, which permits the users to adjust the contrast and segmentation of blood vessels in retinal images.

Extraction and Visualization of Ocular Blood Vessels in 3D Medical Images Based on Geometric Transformation Algorithm

“…Zhang et al used the Kirsch operator as the convolution kernel for matched filtering, but its segmentation detection effect in microscopic blood vessels with low contrast is still not obvious, and there are voids in the middle of the segmented detected blood vessels [14]. Rocha et al used Gabor transform to manipulate fundus images to extract pixels and regions that match the vessel features to obtain segmentation detection results, but the implementation process needs to be supplemented with a large number of presegmented standard images, and in most cases, there are insufficient conditions for implementation [15]. Joseph et al used a simple pulse-coupled neural network and a fast 2D Otsu threshold segmentation method combined with a distributed genetic algorithm to extract the main blood vessels and proposed a method to automatically detect the retinal blood vessels in the fundus [16].…”

Color Feature Segmentation Image for Identification of Cotton Wool Spots on Diabetic Retinopathy Fundus

“…CLAHE technique is used to increase image contrast by providing a limit value (clip limit) so that the image looks clearer [23]. The contrast value given must be limited so that there is no excessive contrast increase in the histogram [24]. To calculate the clip limit of a histogram using Formula (2).…”

Localizing Retinal Blood Vessels In Fundus Images using Fuzzy Logic Approach

“…According to Marin et al [66] experiments, the method suggested by Ricci et The main issue with supervised methods is their dependency on robust feature extraction to classify the pixels of the image into vessels and background. Since this feature vector must be calculated for each pixel, the algorithms become almost complicated, and the procedure will be time-consuming due to the training stage [47,48,68].…”

“…Da Rocha et al [48] performed pre-processing on the green channel of the RGB images from the DRIVE database by creating the complement of the image and making a mask of the region of interest (ROI) of the retina area, followed by an adaptive method of histogram equalization. The next step consists of 2D Gabor Wavelet Transform (GWT) and closing operation.…”

“…These methods can be divided into two groups: unsupervised and supervised methods. Unsupervised methods apply different image processing filters, mathematical morphology and vessel tracking methods to extract blood vessels [20,26,31,38,45,47,48,53,57,65,68]. On the other hand, the supervised methods apply pixel classification techniques to categorize objects in the image as vessels and non-vessels [41,43,49,50,51,52].…”