KARDIYAK ARITMILERIN TANISI İÇIN FARKLI TEMSILLER ALTINDA ECG SINYALININ SINIFLANDIRILMASI

Diagnosis of cardiac arrhythmias plays an important role in the clinical stage. Biomedical engineering solutions focus on Electrocardiographic (ECG) signal classification for arrhythmia detection. The nature of ECG signals with a rhythmic structure raises a question: whether the cyclic transformation of the signal may enhance classification success. With this motivation, our study comprises cyclic transforming of the ECG signal, testing the cyclic structure, extracting capable features that represent signal characteristics, and distinguishing healthy or arrhythmia points as well as specific arrhythmia subjects, in various classification methods, and finally evaluating and comparing the performances according to the accuracy metric. Within this scope, outcomes reveal that the ECG signal can be represented with a cyclic formation efficiently, feature sets that are organized with classical statistical, wavelet, circular statistical, circular distance-based, and regression-based methods provide a comprehensive perspective. The accuracy results of the classification methods provide several analyses that are based on signal duration, data representation type, feature set contribution, and classification strategy, which together offer clear and practical information for understanding arrhythmia classification during the clinical evaluation stage.