Variational Modal Decomposition Joint Independent Component Analysis of PPG Noise Reduction Algorithm for Blood Oxygen Calculation.

The detection of blood oxygen saturation (SpO2) is a critical detection item for human physiological parameters in clinical medicine. The blood oxygen monitoring method based on the photoplethysmogram (PPG) signal is easy to operate. It does not cause trauma to the subject and is gradually becoming the mainstream technology for blood oxygen saturation detection. This study proposes a noise reduction algorithm based on the PPG signal to improve the accuracy of the calculation of blood oxygen saturation. The algorithm reconstructs the noise reference signal through the modes decomposed by Variational Modal Decomposition (VMD). It then constructs a two-dimensional signal from the noisy reference signal and the original signal, which is processed by the Independent Component Analysis (ICA) algorithm to obtain the noise-reduced signal. The correlation coefficient between the noise-reduced signal and the original signal reaches 0.934. Compared with other noise reduction algorithms, the signal-to-noise ratio of this algorithm is improved by 2.1 dB, 1.49 dB, and 0.78 dB, respectively. At the same time, the improvement of root-mean-square error reached 48.26%, 34.55% and 1,83% respectively. From the results of the standard deviation of SpO2 affected by noise, the result range of the VMD-ICA algorithm is 6.72-7.11, which is the lowest value and the smallest fluctuation range compared with the other three algorithms, showing relatively the best stability and accuracy. These results fully demonstrate that the noise reduction algorithm combined with VMD and ICA has high accuracy and stability when processing PPG signals and effectively improves the quality of PPG signals and the accuracy of blood oxygen calculations.