Resumen
Sudden cardiac death (SCD) is a global threat that demands our attention and research. Statistics show that 50% of cardiac deaths are sudden cardiac death. Therefore, early cardiac arrhythmia detection may lead to timely and proper treatment, saving lives. We proposed a less complex, fast, and more efficient algorithm that quickly and accurately detects heart abnormalities. Firstly, we carefully examined 23 ECG signals of the patients who died from SCD to detect their arrhythmias. Then, we trained a deep learning model to auto-detect and distinguish the most lethal arrhythmias in SCD: Ventricular Tachycardia (VT) and Ventricular Fibrillation (VF), from Normal Sinus Rhythm (NSR). Our work combined two techniques: Wavelet Transform (WT) and pre-trained Convolutional Neural Network (CNN). WT was used to convert an ECG signal into scalogram and CNN for features extraction and arrhythmias classification. When examined in the MIT-BIH Normal Sinus Rhythm, MIT-BIH Malignant Ventricular Ectopy, and Creighton University Ventricular Tachyarrhythmia databases, the proposed methodology obtained an accuracy of 98.7% and an F-score of 0.9867, despite being less expensive and simple to execute.