Título: Detecção de Doenças Cardíacas Através de Lifting Wavelet e Redes Neurais Artificiais
Autores: Côco, Klaus Fabian; Almeida, Ailson Rosetti; Salles, Evandro Ottoni Teatini
Resumo: Previous works show that the energy spectrum, taken from the coefficients of each level of the wavelet decomposition of ECG signals, forms good feature vectors for the classification of a number of specific heart anomalies. The required processing is simple. On the other hand, it is highly desirable (and also required by some cardiologic analyses) the ECG recording during the normal daily patient’s activities. We are also interested in real time patient monitoring for the detection and alarm of some critical situation. For such, we need a portable, inexpensive, and computationally efficient hardware/software system, leading to the use of readily available Palmtop computer augmented with the instrumentation hardware and embedded software. The present study exploits the use of lifting wavelets (also called second generation wavelets) as a feature extractor engine, followed by a neural network classifier. The purpose of this work is the development of such a classification system, with emphasis for efficiency and portability.
Código DOI: 10.21528/CBRN2003-052
Artigo em PDF: 6CBRN_052.PDF
Arquivo BibTex: 6CBRN_052.bib