Metodologia para localização de ciclo cardíaco a partir do sinal eletrocardiográfico

Abstract

This dissertation proposes a cardiac cycle detection algorithm based on digital signals generated by ECG (electrocardiogram) devices. Detection is made through a threshold, seeking R wave occurrences, as represented by its second derivate. A threshold is established dynamically based on the last 10 seconds of the signal and its value is calculated taking the standard deviation of the second derivate in the period into account. The derivates are applied to the filtered ECG signal, using a moving average filter (20 samples). To test the model performance operation, a software application was developed prepared to receive files containing digitalized data, in 16-bit words with up to 12 derivates as input and running at a configurable acquisition frequency. The application can receive real-time data acquired directly from the ECG device with digital output channels. The software that has been developed provides statistical data and information of interest related to the analyzed ECG. The algorithm was tested through experimentation with hundreds of samples coming from normal patients and from people suffering from cardiopathies. The accuracy level was above 99%.