ECG arrhythmia classification by using a recurrence plot and convolutional neural network

• A new ECG arrhythmia classification method combining recurrence plot (RP) and deep learning in two stages is proposed. 1D signals were converted into 2D images using the RP tool to expose features for CNN classify. We applied since leverages spatial information, therefore well suited image classification. To archive better results, we took 2-second segments, then a two-stage R-peak detection. Cardiovascular diseases affect approximately 50 million people worldwide; thus, heart disease prevention one of most important tasks any health care system. Despite high popularity electrocardiography, superior automatic electrocardiography (ECG) signal analysis methods are required. The aim this research was design effectively classifying by segments signals. In first stage, noise ventricular fibrillation (VF) categories distinguished. second atrial (AF), normal, premature AF, VF Models trained tested databases publicly available at website PhysioNet. MIT-BIH Arrhythmia Database, Creighton University Ventricular Tachyarrhythmia Atrial Fibrillation Malignant Ectopy Database used compare six types arrhythmia. Testing accuracies up 95.3 % ± 1.27 98.41 0.11 achieved detection respectively, after five-fold cross-validation. conclusion, study provides clinicians with an advanced methodology detecting discriminating between different types.