No abstract

Segmentation of retinal blood vessels is the first step for several computer aided-diagnosis systems (CAD), not only for ocular disease diagnosis such as diabetic retinopathy (DR) but also of non-ocular disease, such as hypertension, stroke and cardiovascular diseases. In this paper, a supervised learning-based method, using a multi-layer perceptron neural network and carefully selected vector of features, is proposed. In particular, for each pixel of a retinal fundus image, we construct a 24-D feature vector, encoding information on the local intensity, morphology transformation, principal moments of phase congruency, Hessian, and difference of Gaussian values. A post-processing technique depending on mathematical morphological operators is used to optimise the segmentation. Moreover, the selected feature vector succeeded in outfitting the symmetric features that provided the final blood vessel probability as a binary map image. The proposed method is tested on three known datasets: Digital Retinal Image for Extraction (DRIVE), Structure Analysis of the Retina (STARE), and CHASED_DB1 datasets. The experimental results, both visual and quantitative, testify to the robustness of the proposed method. This proposed method achieved 0.9607, 0.7542, and 0.9843 in DRIVE, 0.9632, 0.7806, and 0.9825 on STARE, 0.9577, 0.7585 and 0.9846 in CHASE_DB1, with respectable accuracy, sensitivity, and specificity performance metrics. Furthermore, they testify that the method is superior to seven similar state-of-the-art methods.

The demand for virtual machine requests has increased recently due to the growing number of users and applications. Therefore, virtual machine placement (VMP) is now critical for the provision of efficient resource management in cloud data centers. The VMP process considers the placement of a set of virtual machines onto a set of physical machines, in accordance with a set of criteria. The optimal solution for multi-objective VMP can be determined by using a fitness function that combines the objectives. This paper proposes a novel model to enhance the performance of the VMP decision-making process. Placement decisions are made based on a fitness function that combines three criteria: placement time, power consumption, and resource wastage. The proposed model aims to satisfy minimum values for the three objectives for placement onto all available physical machines. To optimize the VMP solution, the proposed fitness function was implemented using three optimization algorithms: particle swarm optimization with Lévy flight (PSOLF), flower pollination optimization (FPO), and a proposed hybrid algorithm (HPSOLF-FPO). Each algorithm was tested experimentally. The results of the comparative study between the three algorithms show that the hybrid algorithm has the strongest performance. Moreover, the proposed algorithm was tested against the bin packing best fit strategy. The results show that the proposed algorithm outperforms the best fit strategy in total server utilization.

Diabetic retinopathy (DR) and glaucoma can both be incurable if they are not detected early enough. Therefore, ophthalmologists worldwide are striving to detect them by personally screening retinal fundus images. However, this procedure is not only tedious, subjective, and labor-intensive, but also error-prone. Worse yet, it may not even be attainable in some countries where ophthalmologists are in short supply. A practical solution to this complicated problem is a computer-aided diagnosis (CAD) system—the objective of this work. We propose an accurate system to detect at once any of the two diseases from retinal fundus images. The accuracy stems from two factors. First, we calculate a large set of hybrid features belonging to three groups: first-order statistics (FOS), higher-order statistics (HOS), and histogram of oriented gradient (HOG). Then, these features are skillfully reduced using a genetic algorithm scheme that selects only the most relevant and significant of them. Finally, the selected features are fed to a classifier to detect one of three classes: DR, glaucoma, or normal. Four classifiers are tested for this job: decision tree (DT), naive Bayes (NB), k-nearest neighbor (kNN), and linear discriminant analysis (LDA). The experimental work, conducted on three publicly available datasets, two of them merged into one, shows impressive performance in terms of four standard classification metrics, each computed using k-fold crossvalidation for added credibility. The highest accuracy has been provided by DT—96.67% for DR, 100% for glaucoma, and 96.67% for normal.

Cancer is a deadly disease that occurs due to rapid and uncontrolled cell growth. In this article, a machine learning (ML) algorithm is proposed to diagnose different cancer diseases from big data. The algorithm comprises a two-stage hybrid feature selection. In the first stage, an overall ranker is initiated to combine the results of three filter-based feature evaluation methods, namely, chi-squared, F-statistic, and mutual information (MI). The features are then ordered according to this combination. In the second stage, the modified wrapper-based sequential forward selection is utilized to discover the optimal feature subset, using ML models such as support vector machine (SVM), decision tree (DT), random forest (RF), and K-nearest neighbor (KNN) classifiers. To examine the proposed algorithm, many tests have been carried out on four cancerous microarray datasets, employing in the process 10-fold cross-validation and hyperparameter tuning. The performance of the algorithm is evaluated by calculating the diagnostic accuracy. The results indicate that for the leukemia dataset, both SVM and KNN models register the highest accuracy at 100% using only 5 features. For the ovarian cancer dataset, the SVM model achieves the highest accuracy at 100% using only 6 features. For the small round blue cell tumor (SRBCT) dataset, the SVM model also achieves the highest accuracy at 100% using only 8 features. For the lung cancer dataset, the SVM model also achieves the highest accuracy at 99.57% using 19 features. By comparing with other algorithms, the results obtained from the proposed algorithm are superior in terms of the number of selected features and diagnostic accuracy.