Currently, there is an increasing interest for setting up medical systems that can screen a large number of people for sight threatening diseases, such as diabetic retinopathy. This paper presents a method for automated identification of exudate pathologies in retinopathy images based on computational intelligence techniques. The color retinal images are segmented using fuzzy c-means clustering following some preprocessing steps, i.e., color normalization and contrast enhancement. The entire segmented images establish a dataset of regions. To classify these segmented regions into exudates and nonexudates, a set of initial features such as color, size, edge strength, and texture are extracted. A genetic-based algorithm is used to rank the features and identify the subset that gives the best classification results. The selected feature vectors are then classified using a multilayer neural network classifier. The algorithm was implemented using a large image dataset consisting of 300 manually labeled retinal images, and could identify affected retinal images with 96.0% sensitivity while it recognized 94.6% of the normal images, i.e., the specificity. Moreover, the proposed scheme illustrated an accuracy including 93.5% sensitivity and 92.1% predictivity for identification of retinal exudates at the pixel level.

No abstract

Detecting blood vessels is a vital task in retinal image analysis. The task is more challenging with the presence of bright and dark lesions in retinal images. Here, a method is proposed to detect vessels in both normal and abnormal retinal fundus images based on their linear features. First, the negative impact of bright lesions is reduced by using K-means segmentation in a perceptive space. Then, a multi-scale line operator is utilized to detect vessels while ignoring some of the dark lesions, which have intensity structures different from the line-shaped vessels in the retina. The proposed algorithm is tested on two publicly available STARE and DRIVE databases. The performance of the method is measured by calculating the area under the receiver operating characteristic curve and the segmentation accuracy. The proposed method achieves 0.9483 and 0.9387 localization accuracy against STARE and DRIVE respectively.

No abstract

In the semantic web, ontology plays an important role to provide formal definitions of concepts and relationships. Therefore, communicating similar ontologies becomes essential to provide ontologies interpretability and extendibility. Thus, it is inevitable to have similar but not the same ontologies in a particular domain since there might be several definitions for a given concept. This paper presents a method to combine similarity measures of different categories without having ontology instances or any user feedback in regard with alignment of two given ontologies. To align different ontologies efficiently, K Nearest Neighbor (KNN), Support Vector Machine (SVM), Decision Tree (DT) and AdaBoost classifiers are investigated. Each classifier is optimized based on the lower cost and better classification rate. Experimental results demonstrate that the F-measure criterion improves up to 99% using feature selection and combination of AdaBoost and DT classifiers, which is highly comparable, and outperforms the previous reported F-measures.