POSSIBILITIES OF USING NEURAL NETWORKS FOR ECG CLASSIFICATION

Some characteristics of the neural network approach have been tested a nd validated for the particular problem of diagnostic classification i n the field of computerized electrocardiography. Two different databas es have been used for the evaluation process: CORDA, developed by the Medical Informatics Department of the University of Leuven, and ECG-UC L, developed by the Cliniques Universitaires Saint-Luc, Universite Cat holique de Louvain. Electrocardiographic signals classified on the bas is of electrocardiographic independent clinical data, with a single di agnosis and no conduction abnormalities, have been considered. Seven d iagnostic classes have been taken into account, including the differen t locations of ventricular hypertrophy and myocardial infarction. Two architectures of neural networks have been analyzed in detail consider ing three aspects: the normalization process, pruning techniques, and fuzzy preprocessing by the use of radial basis functions. The comparis on of the results obtained with the two databases will be discussed in detail.