基于光电容积脉搏波的血红蛋白无创检测技术

Noninvasive Detection of Hemoglobin Based on Photoplethysmography

提出一种基于多波长光电容积脉搏波(PPG)的检测技术,实现人体总血红蛋白(tHb)浓度的无创测量。首先,根据改进的Lambert-Beer定律,推导tHb浓度与PPG信号的理论关系。然后,设计无创检测系统,采集6种特定波长光源照射下的PPG信号。最后,提取信号的幅度特征,分别应用偏最小二乘回归(PLSR)和支持向量回归(SVR)算法预测tHb浓度。实验结果表明,两种回归模型的预测值与标准值均具有较高的相关性(r>0.7)。SVR模型预测的结果更准确,在82例样本上的预测值与标准值的平均绝对误差为0.849 8 g/dL,相关系数为0.845 7。进一步证实了无创检测tHb浓度的可行性,提供了具有临床实用性的实现方案。

To determine the total hemoglobin(tHb)concentration noninvasively,a measurement based on multi-wavelength photoplethys-mography(PPG)is proposed.The theoretical relationship between tHb content and PPG signals is derived according to modified Lambert-Beer’s Law.Then a noninvasive measurement system is conducted to collect PPG signals from fingertip irradiated by 6 kinds of specific wavelength light sources.Finally,the amplitude features of these signals are extracted,and Partial Least Squares Regression(PLSR)and Support Vector Regression(SVR)algorithms are implemented to predict tHb concentration.Experimental results manifest that the predicted values of both regression models are highly correlated with the standard values(r>0.7).SVR achieves better performance,the mean absolute error between predicted values and standard values on all 82 samples is 0.8498 g/dl,and the correlation coefficient is 0.8457.This study further confirmed the feasibility of noninvasive detection of tHb concentration and provided a clinically practical prototype.