A morphological approach to pulse feature extraction from the digital volume pulse

The digital volume pulse (DVP) is a physiological signal obtained from the photoplethysmography. Since the photoplethysmography measures volumetric changes of blood occurred by heart contraction and properties of blood vessels, DVP contains several feature parameters that reflect cardiovascular events, such as arterial stiffness. In order to extract these parameters, a number of algorithms have been proposed and applied in various clinical settings. However, these algorithms mainly focus on the feature extraction based on the mathematical representation of the DVP. This approach may show high performance in regular pulse waveform, but there is a limit to apply these algorithms to the DVP that includes various pulse shapes. In order to cope with this problem, we have proposed a morphological algorithm based on the knowledge of human recognition way. From the result of the performance evaluation, we have achieved the low false positive rate of 7.41% and the low false negative rate of 1.91%. With such low error rate and increased applicability to various pulse waveforms, the proposed algorithm can be utilized as an effective solution for the DVP-based cardiovascular function assessment.