Zafer Yavuz

Retinal blood vessel segmentation using gabor filter and top-hat transform

Retinal blood vessel segmentation is a widely used process in diagnosis of various diseases such as diabetic retinopathy, glaucoma and arteriosclerosis. Therefore, an automated tool developed for vessel segmentation could be employed in diagnosis of those illnesses to help ophthalmologists. In this paper, we suggest a method to segment retinal blood vessels automatically. In the method, we apply top-hat transform after Gabor filter to enhance blood vessels. Later on, the output of the transformation is converted to binary image with p-tile thresholding. In order to test the developed system 20 images obtained from STARE database are used for performance evaluation. The results shows 86.31% of true positive rate (sensivity) and 92.90% of accuracy, which is promising.

Blood Vessel Extraction in Color Retinal Fundus Images with Enhancement Filtering and Unsupervised Classification

Retinal blood vessels have a significant role in the diagnosis and treatment of various retinal diseases such as diabetic retinopathy, glaucoma, arteriosclerosis, and hypertension. For this reason, retinal vasculature extraction is important in order to help specialists for the diagnosis and treatment of systematic diseases. In this paper, a novel approach is developed to extract retinal blood vessel network. Our method comprises four stages: (1) preprocessing stage in order to prepare dataset for segmentation; (2) an enhancement procedure including Gabor, Frangi, and Gauss filters obtained separately before a top-hat transform; (3) a hard and soft clustering stage which includes K-means and Fuzzy C-means (FCM) in order to get binary vessel map; and (4) a postprocessing step which removes falsely segmented isolated regions. The method is tested on color retinal images obtained from STARE and DRIVE databases which are available online. As a result, Gabor filter followed by K-means clustering method achieves 95.94% and 95.71% of accuracy for STARE and DRIVE databases, respectively, which are acceptable for diagnosis systems.

Retinal fundus image registration using bifurcation and crossover points

Computer Aided Diagnosis (CADx) systems have been reached to a large usuage area in recent years. Blood vessels extracted from retinal fundus images provide us important features for diagnosis and treatment of some systematic diseases. These features such as bifurcation and crossover points are mostly used in image registration applications. In this paper, an image registration study is performed by using binary retinal vessel map. Firstly, a thinnig operation is performed in order to obtain skeletonized vessel image. Then feature points like bifurcations and crossovers are extracted in order to use for image registration. Lastly, a characteristic matrix is extracted to represent the retinal vessel network. Similar operations are performed on a second rotated, translated and scaled retinal vessel network image. By this way, two retinal vessel network images belong to the same person are compared and each feature points are paired with each others in order to develope an image registration process.

Vessel segmentation in retinal images using multiscale image enhancement and clustering

Some diseases in human body such as diabet could be affect the morphology of the retina. The diagnosis and treatment of these diseases can be made easily by improved computerized techniques. Retinal blood vessel segmentation phase is an important step for diagnosis and treatment. Blood vessel segmentation in color retinal fundus images is employeed in this paper. First, a preprocessing step is performed and then multiscale Frangi filter is applied in order to enhance blood vessels. Afterwards Fuzzy C-means clustering method is used to obtain binary vessel image. Finally, a postprocessing step is performed to increase performance.We use two publicly available retinal fundus image databases STARE and DRIVE to measure the performance of the system. As a result we get 95.95% of accuracy for STARE database and 95.95% of accuracy for DRIVE database.

Blood vessel segmentation from retinal images based on enhencement methods

Usage of Computer Aided Diagnostic (CAD) systems is increasing rapidly. Blood vessel segmentation on retinal fundus images could be used as a CAD system for diagnosis of various retinal diseases. In this paper, blood vessels are segmented on retinal fundus images. Firstly, Gabor filter and morphological top-hat transform are applied after preprocessing step in order to enhance blood vessels. Afterward, we performed p-tile thresholding method to obtain binary vessel image. At the last step a post processing method is applied to increase accuracy. In order to test the developed system, the images obtained from STARE and DRIVE databases are used. Finally, 94.02% of accuracy for STARE database and 94.59% of accuracy for DRIVE database are obtained as a result, which is promising.