ABP peak detection using energy analysis technique

Abstract

This paper describes an algorithm for the automatic detection of systolic peaks in the Arterial Blood Pressure (ABP) signal. ABP waveform is rich in pathological information such as heart rate, systolic, mean and diastolic pressure therefore it is very important for clinical diagnosis in cardiology. The proposed algorithm deals with the development of energy based technique for the detection of systolic peaks in ABP signal that is of great importance in evaluating heart rate computation and serves as the basis for the extraction of other ABP features. The algorithm has been developed based on the window based energy analysis of ABP signal. This energy domain offers an easy interpretation of the ABP signal for the detection of systolic peaks in ABP signal. The ABP signal under test undergoes window based energy analysis by selecting a window of 100 ms duration. The window is shifted along time axis and the energy of samples falling in the succeeding window is calculated. The areas in ABP signal where systolic peaks are available appear as high energy zones. Window based amplitude threshold and interval threshold are applied to reject the unwanted peaks. The algorithm is easier to implement just after removal of the low frequency samples in ABP signal. ABP signals are downloaded from MGH/MF waveform database and algorithm has been developed on five minute segment of ABP signals. The method has been validated based on beat annotations of five minute segment of corresponding ECG signal from modified lead II.