Enhancing ECG Diagnosis Using Hybrid Automated Technique

The electrocardiogram (ECG) is a test of electrical activities of the heart. To detect cardiac conditions different detection techniques are used. In this paper, a novel hybrid system combining a modified scaling technique and Wavelet technique is implemented. It is applied to enhance the accuracy of filtration, denoising and diagnosis techniques. In previous computerized diagnosis techniques, either filtration or denoising is used. However, in this system, filtration and denoising are mixed in pre-processing to give a pure signal. This research deems as the premier work to utilize, in the diagnosis phase, the time feature of each wave and its location in the ECG signal. In contrast to previous automated techniques, the proposed hybrid system is based on three factors to detect and diagnose the ECG episodes; namely amplitude, frequency and time location scaling of the ECG signal. Mixing effectively these three factors in the diagnosis phase allows the detection of more episodes, gives more accurate and faster results. As the results demonstrate, the previous computerized techniques' average detection accuracy does not exceed 80 %, while the proposed hybrid technique average accuracy overtakes 97% with a good average time consumption equal to 0.05 seconds. Furthermore, the proposed system overcomes some of the previous challenges and detects more new episodes that have never been diagnosed before by any automated systems. This system can help the cardiologists to take more confident decisions in their diagnoses.