An innovative dicrotic notch detection algorithm which combines rule-based logic with digital signal processing techniques.

Abstract

Automated, real-time localization of the dicrotic notch, a component of the arterial pressure waveform, represents a deceptively complex problem in computerized biomedical signal processing. The high-frequency nature of the notch can make it difficult to distinguish from artifactual noise or from other high-frequency physiological components of the waveform. In addition, the contour of the notch varies with vascular status and with propagation through arterial beds, requiring any detection algorithm to recognize various possible notch conformations. Finally, location of the notch along the waveform may vary widely depending on other hemodynamic variables, further complicating detection algorithms. We have reviewed various published algorithms and have implemented a number of them to determine the strengths and shortcomings of each. We then developed a reliable and accurate hybrid algorithm which utilizes the strengths of the various algorithmic approaches reviewed; after analyzing the waveform, the algorithm selects the most appropriate method for accurate notch localization based on a series of waveform features. The application of rule-based logic represents a relatively unique approach to digital signal processing.