Ant Colony Cooperative Strategy in Electrocardiogram and Electroencephalogram Data Clustering
Cooperation in natural processes is very important feature, which is modeled by many nature-inspired algorithms. Nature inspired metaheuristics have interesting stochastic properties which make them suitable for use in data mining, data clustering and oth
- PDF / 202,353 Bytes
- 11 Pages / 439.324 x 666.138 pts Page_size
- 23 Downloads / 207 Views
Abstract Cooperation in natural processes is very important feature, which is modeled by many nature-inspired algorithms. Nature inspired metaheuristics have interesting stochastic properties which make them suitable for use in data mining, data clustering and other computationally demanding application areas. It is because they often produce robust solutions in fairly reasonable time. This paper presents an application of clustering method inspired by the behavior of real ants in the nature in biomedical signal processing. The ants cooperatively maintain and evolve a pheromone matrix which is used to select features. The main aim of this study was to design and develop a combination of feature extraction and classification methods for automatic recognition of significant structure in biological signal recordings. The method is targeted towards speeding up and increasing objectivity of identification of important classes and may be used for online classification. Inherent properties of the method make it suitable for analysis of newly incoming data. The method can be also used in the expert classification process. We have obtained significant results in electrocardiogram and electroencephalogram recordings, which justify the use of such method.
1 Introduction Biological signals, for example electroencephalogram and electrocardiogram, contain similar structures, which have to be discovered within the data and furthermore evaluated. With the oncoming boost in personal medical electronics and portable monitoring technology, there is a growing trend in the amount of data that must be processed and evaluated by the physicians. M. Bursa and L. Lhotska Czech Technical University in Prague, Technicka 2, 166 27 Prague 6, Czech Republic; BioDat Research Group (http://bio.felk.cvut.cz) {bursam, lhotska}@fel.cvut.cz M. Bursa and L. Lhotska: Ant Colony Cooperative Strategy in Electrocardiogram and Electroencephalogram Data Clustering, Studies in Computational Intelligence (SCI) 129, 323–333 (2008) c Springer-Verlag Berlin Heidelberg 2008 www.springerlink.com
324
M. Bursa, L. Lhotska
This study aims at design and automated development of classification models for automatic recognition of important patterns in biological signal recordings. The doctors often work under stress conditions (time stress and fatigue) and the error rate of the human expert increases when working under stress. Therefore automated methods are developed, which aim at speeding up and increasing objectivity of identification of relevant classes and may be used for online classification. However, results of any automated method should be provided only as a hint to the doctor, as they do not consider many other aspects (medication, diagnosis, treatment, patient medical history, etc.). The final decision is to be made by qualified physician only. Biological signal processing workflow consists of the following main processes: signal pre-processing, signal transfer and/or storage, digital signal processing and feature extraction, clustering of the similar data (mainly
Data Loading...