Maximal Overlap Discrete Wavelet Packet Transform for ECG Signal-Based Cardiac Ailments Classification

The critical identification and subsequent therapeutic treatment of cardiac diseases requires a correct categorization of ECG signals. The presented feature of separating data is highly valuable for diagnosing and treating heart diseases. A new methodology for feature extraction from ECG signals is presented in the current contribution. In this approach, Maximal Overlap Discrete Wavelet Packet Transform (MODWPT) is employed. This approach was formerly applied for the purpose of improving detecting of fine alterations in rhythms of the heart. Once the features are procured, they are employed with the use of machine learning strategies to detect those features. Furthermore, a high degree of specificity is achieved in distinguishing several cardiac conditions such as arrhythmias and myocardial infarction amongst others, using this process. There exists another possibility to increase the stability of the method – factorization of the signal, that can help to eliminate the noise. About this method, one can speak about the possibility of automation of the diagnostic process, as an economical tool having reliable indicators for doctors. There is a very good deal of potential that this method provides.