Adaptive Pulse Wave Extraction Method for Radial Artery Video

Pulse waves play an important role in the diagnosis and treatment of cardiovascular diseases and Chinese medicine. Today, the accuracy of noncontact pulse wave detection still needs to be improved. In this article, the Eulerian amplification principle is used to amplify the pulse motion part of the sampled radial artery video to visualize the small movements caused by the blood impinging on the skin. We apply an interframe windowing differential strategy to select the region of interest (ROI). Segmented baseline cancellation removes very high and very low trends from the original waveform. The spike filter function removes wild point values in the pulse waveform by sliding window mean comparison, and wavelet transform filtering combined with locally weighted regression filtering effectively filters out high-frequency noise in the pulse waveform signal. The singular spectrum is decomposed and reconstructed to remove the burr-like noise in the pulse wave and to preserve the effective details of the signal as much as possible. The Euler amplification factor and reconstruction factor are adaptively adjusted to enhance the accuracy and robustness of pulse wave extraction. From the experimental results, the mean square error of 0.1999 and the standard deviation of 0.2700 are the smallest values in the comparison experiments, indicating that this experiment is feasible and effective in increasing the energy ratio and reducing the pulse wave measurement error. The Pearson product-moment correlation coefficient (PPMCC) is 0.6713, which indicates that the waveform measured in this experiment has a good fit with the contact pulse wave.