Detection of ECG Arrhythmia Using Large Choi Williams Time-Frequency Feature Set


Medical Technologies National Congress (TIPTEKNO), Trabzon, Türkiye, 12 - 14 Ekim 2017 identifier identifier


Early detection and monitoring of heart diseases increase human quality of life and this can prevent negative consequences. It is even more important because it can prevent sudden deaths. In today's technology, these operations can be done with telemedicine systems. In this work, appropriate methods have been proposed for telemedicine systems. The proposed system is of two classes and is based on detection of arrhythmia from healthy and diseased ECG signals. MIT-BIH Arrhythmia database was used in the study. A total of 103026 R-R interval were used in this database. In this study, the Choi-Williams transformation is used as an feature extraction method. The performance results are given as accuracy, specificity and positive predictive accuracy, respectively 94.67%, 94.97%, 92.57%, 97.36%, 97.23%