Classification of Heart Blockage Using Particle Swarm Optimization-Based Probabilistic Neural Network

Around half of all patients suffer from sudden cardiac death and hence heart blockage is still the primary cause of death worldwide. The early identification and prediction of these cardiac diseases have emerged as a major study area and it is critical for the survival of cardiac patients. Electrocardiography (ECG) is regarded as a foremost screening approach for patients with heart problems thereby contributing to the reliable prediction of cardiac diseases. This paper exploits the signal processing of ECG signals with four levels like preprocessing, extraction of feature categorization and segmentation. Initially, in the preprocessing stage, Notch filter is used to eliminate noise and the frequency is maintained in a definite band. After the removal of noises, the signal is divided into several segments by utilizing Karl Pearson’s coefficient of correlation. Subsequently, the desired features of each segment are extracted through the implementation of genetic algorithm. The extracted features are subjected to classification by the particle swarm-optimized neural network (PSNN), which compares the trained data with the test data to determine the accurate blockage.