基于试验方差分析的血氧测量波长选取的研究

发布时间：2018-12-23 12:11

【摘要】：为了提高血氧饱和度光谱测量的精度和可靠性,增强血氧饱和度测量数据的有效性,提出了基于试验方差分析的血氧测量波长选取的方法。通过分析不同波长组合对应的血氧饱和度系数分布情况,合理地利用统计理论,从中选取较好的波长组合。实验中以不同血氧饱和度临床数据为基础,建立了不同波长组合(660和940nm,660和805nm,805和940nm)计算出的血氧饱和度系数的单因素试验方差分析模型,参照分析模型中的F指标和p参数,比较分析不同波长组合的显著性,进而从光谱数据中选取出相对较好的波长组合,为进一步建模分析提供了可靠的中间数据。结果表明,选取660和805nm的波长组合引入的总误差相对较小,能更显著的代表血氧饱和度的变化,相比其他的波长组合进行分析能够提高血氧测量精度。本研究将试验方差分析方法用于血氧测量中波长组合的优选,效果显著,为血氧饱和度的测量和其他相关光谱的特征波长选取以及定量分析提供了新思路。试验方差分析方法有助于从光谱中提取代表被测量的有效信息。
[Abstract]:In order to improve the accuracy and reliability of blood oxygen saturation spectrum measurement and enhance the validity of blood oxygen saturation measurement data, a method of wavelength selection for blood oxygen saturation measurement based on test variance analysis was proposed. By analyzing the distribution of blood oxygen saturation coefficient corresponding to different wavelength combinations, a better wavelength combination is selected by using statistical theory. Based on the clinical data of different blood oxygen saturation, a single factor analysis of variance model of the coefficient of blood oxygen saturation calculated by different wavelength combinations (660 and 940nmnmt660and 805nmH805 and 940nm) was established. Referring to the F index and p parameter in the analysis model, the significance of different wavelength combinations is compared and analyzed, and a relatively good wavelength combination is selected from the spectral data, which provides reliable intermediate data for further modeling and analysis. The results show that the total error introduced by the wavelength combination of 660 and 805nm is relatively small, and it can represent the change of oxygen saturation more significantly. Compared with other wavelength combinations, the precision of blood oxygen measurement can be improved. In this study, the experimental variance analysis method was applied to the optimal selection of wavelength combination in blood oxygen measurement, which provided a new idea for the measurement of blood oxygen saturation and the selection of characteristic wavelengths of other related spectra as well as the quantitative analysis. The method of variance analysis is helpful to extract the effective information representing the measurement from the spectrum.
【作者单位】：天津大学精密测试技术及仪器国家重点实验室;军事交通学院军用车辆系;天津师范大学物理与电子信息学院;天津大学天津市生物医学检测技术与仪器重点实验室;
【基金】：国家自然科学基金项目(30973964) 天津市应用基础及前沿技术研究计划(11JCZDJC17100) 天津市科技计划项目:科技型中小企业创新基金项目(10ZXCXSY10400)资助
【分类号】：O657.33;R318.6

【参考文献】