Premature Ventricular Contraction Classification Based on Spiral Search - Manta Ray Foraging and Bi-LSTM

Abstract

Cardiovascular Diseases (CVDs) have become a burden to the healthcare system due to the increase in the number of patients and the ratio of mortality. Various techniques are applied for the classification of Premature Ventricular Contraction (PVC) and existing methods have lower sensitivity due to more similarity of features between classes. This research proposes the Spiral Search -Manta Ray Foraging (SP-MRF) algorithm to select the unique features that help to increase the sensitivity of the model. The Convolutional Neural Network (CNN) based model is used for the feature extraction from the input signals and the SP-MRF method selects features from extracted features. The Spiral Search technique increases the exploitation of the search method that helps to distinguish between more similar features and select unique features for classes. The selection of unique features helps to increase the sensitivity and efficiency of the model in PVC classification. The selected features were applied for the Bi-directional Long Short Term Memory (BiLSTM) for PVC classification and evaluated the performance. The efficiency of the SP-MRF-BiLSTM model is evaluated on the MIT-BIH dataset for PVC classification. The SP-MFR-BiLSTM model has 99.97 % accuracy, and 99.52 % sensitivity and the KNN-deep metric has 99.59 % accuracy and 96.7 % sensitivity in the model.

Efficient ECG Beats Classification Techniques for The Cardiac Arrhythmia Detection Based on Wavelet Transformation

“…Most researches focus on identifying and detecting the different types of Arrhythmias by classifying different types of heartbeats, especially premature ventricular contractions (PVCs), considered the most dangerous and important essential that causes fatal arrhythmias [2].…”