Hüseyin Pehlivan

Classification of Premature Ventricular Contraction in ECG

Cardiac arrhythmia is one of the most important indicators of heart disease. Premature ventricular contractions (PVCs) are a common form of cardiac arrhythmia caused by ectopic heartbeats. The detection of PVCs by means of ECG (electrocardiogram) signals is important for the prediction of possible heart failure. This study focuses on the classification of PVC heartbeats from ECG signals and, in particular, on the performance evaluation of time series approaches to the classification of PVC abnormality. Moreover, the performance effects of several dimension reduction approaches were also tested. Experiments were carried out using well-known machine learning methods, including neural networks, k-nearest neighbour, decision trees, and support vector machines. Findings were expressed in terms of accuracy, sensitivity, specificity, and running time for the MIT-BIH Arrhythmia Database. Among the different classification algorithms, the k-NN algorithm achieved the best classification rate. The results demonstrated that the proposed model exhibited higher accuracy rates than those of other works on this topic. According to the experimental results, the proposed approach achieved classification accuracy, sensitivity, and specificity rates of 99.63%, 99.29% and 99.89%, respectively.

Feature selection using genetic algorithms for premature ventricular contraction classification

Cardiac arrhythmia is one of the most important indicators of heart disease. Premature ventricular contractions (PVCs) are a common form of cardiac arrhythmia caused by ectopic heartbeats. The detection of PVCs by means of ECG (electrocardiogram) signals is important for the prediction of possible heart failure. This study focuses on the classification of PVC heartbeats from ECG signals and, in particular, on the performance evaluation of selected features using genetic algorithms (GA) to the classification of PVC arrhythmia. The objective of this study is to apply GA as a feature selection method to select the best feature subset from 200 time series features and to integrate these best features to recognize PVC forms. Neural networks, support vector machines and k-nearest neighbour classification algorithms were used. Findings were expressed in terms of accuracy, sensitivity, and specificity for the MIT-BIH Arrhythmia Database. The results showed that the proposed model achieved higher accuracy rates than those of other works on this topic.

Comparison of classification algorithms in classification of ECG beats by time series

Today one of the most important health problems are fatal heart related diseases. Early diagnosis and treatment of heart disease can prevent sudden death. Detected through the human body and seen as a result of activity of the hearts electrical signals is called electrocardiogram (ECG). ECG signal, which can be easily obtained without causing any harm to patients body, is a good indicator of the disorder during operation of the hearth. In this study, Normal beats (N), left bundle branch block (LBBB), right bundle branch block (RBBB) and Paced beat(P) beats are classified and the classification performance has been analyzed. Time series of the signal is used as an input vector for classification algorithms instead of extracting features from the signal. Independent component analysis (ICA) is used for feature reduction. Neural networks, k-nearest neighbour, Bayes, and Decision trees classification algorithms were used. In this study, kNN showed best accuracy rates.

Score calculation from final tournament tables

In this paper, we deal with the issue of determining the scores of all matches involved in a football tournament. The final table of the tournament, which shows the standings of the teams, is taken as the initial data of the problem. This is a different kind of combinatorial problems which require the construction of valid initial states according to some given final state. We use a rules-based method to solve the problem, analyzing the search space and introducing the notion of black&white graphs. The table data is firstly used to compute possible results of all played matches. Based on the results as well as the total number of scored and conceded goals, possible scores of the matches are then computed. The solution strategy is experimented on several final tables from previous World Cup tournaments. Other experiments are conducted for various team standings, measuring the time required to process some specific data of up to 10 teams.

Tournament scoring problem

In this paper, we describe a new scoring problem for tournament environments, called tournament scoring problem. Unlike classical combinatorial problems with some initial state and constraints, the described problem is concerned with the construction of valid initial states according to some given final state and constraints. The problem can easily be generalised to incorporate various sports disciplines played in both tournament and league environments. Given a football tournament, this involves determining possible scores of all matches between teams, using the final state of the tournament table. We propose a constraint-based approach to a feasible solution of the problem for football tournaments, examining the subject of multi-parameter partition and introducing the notion of black&white graphs. Using some particular table data, experimental results are also presented along with the number of different scores and their computation time.

Analysis of Wind Data, Calculation of Energy Yield Potential, and Micrositing Application with WAsP

The parameters required for building a wind power plant have been calculated using the fuzzy logic method by means of Wind Atlas Analysis and Application Program (WAsP) in this study. Overall objectives of the program include analysis of raw data, evaluation of wind and climate, construction of a wind atlas, and estimation of wind power potential. With the analysis performed in the application, the average wind velocity, average power density, energy potential from micrositing, capacity factor, unit cost price, and period of redemption have been calculated, which are needed by the project developer during the decision-making stage and intended to be used as the input unit in the fuzzy logic-based system designed. It is aimed at processing the parameters calculated by the designed fuzzy logic-based decision-making system at the rule base and generating a compatibility factor that will allow for making the final decision in building wind power plants.

Vessel segmentation in angiography images with multi-scale filters

Analysis of coronary angiography is very important for the diagnosis and treatment of heart diseases. The quality of the image is very low due to lack of usage of the X-ray and injected material, which are harmful to human health. For accurate segmentation and analysis the images need to be enhancement. In this study, a Gabor and 2d comparison filter was used to enhance the images and to highlight the vessel structures. It calculates the responses of different size and directional filters in both filters used. The improved images were segmented by P-Tile and Otsu thresholding methods and their performance was compared. The segmentation was performed on 5 different angiograms. The Gabor filter and the P-Tile thresholding method were found to be more successful with accuracy rate of mean all experiment %90,86 and maxium 92.4%.

Solution of limit problems with symbolic computation

In this paper, the solution of limit problems, which is an important subject of high school and university mathematics is presented by using JavaCC code generation tool and symbolic computation methods. Although JavaCC is generally used for generating programming language interpreters, in a similar way it can also be used in the evaluation of mathematical expressions. In this work, first the general grammar rules of limit expressions is extracted. Then parser code for the limit expressions is generated with JavaCC according to the grammar rules. Using the list of the tokens into which a limit expression is parsed with this code, an Abstract Syntax Tree (AST) is constructed. Finally, the solution is obtained by interpreting the AST with a class of Visitor Design Pattern. The study can be regarded as a promising contribution to computer assisted education.