胆红素无创光学检测相关基础问题的研究

发布时间：2018-01-31 00:00

本文关键词： 胆红素无创光学检测血红蛋白浮动基准理论蒙特卡洛模拟　出处：《天津大学》2012年硕士论文　论文类型：学位论文

【摘要】：胆红素代谢异常会引起一系列的疾病,严重地危害着人们的健康,因此,临床上把血清胆红素浓度作为日常肝、胆功能检查中的一项重要生理指标。目前临床人体胆红素浓度检测方法主要为有创的血清胆红素检测,这种方法需要取血,会对患者造成疼痛不适感和感染的风险,且不能进行实时地胆红素浓度监测。无创光学胆红素检测方法以其无创、快速、简单等特点在临床胆红素检测中发挥了重要作用,但目前市场上应用无创光学胆红素检测方法的产品如经皮胆红素监测仪等普遍精度不高,只能用于新生儿黄疸疾病的初次筛查,妨碍了产品的推广使用。为了使用浮动基准理论提高胆红素检测的精度,研究了无创光学胆红素检测的特点,通过计算和总结胆红素检测相关的各项皮肤模型光学参数,在可见光波段特定检测波长下进行蒙特卡洛模拟,得到随胆红素浓度变化不同径向位置扩散反射光的分布特性,发现未能找到胆红素检测的基准位置,这是由于胆红素在可见光波段具有强吸收的独特光学性质引起的;随后基于纯吸收介质中的基准波长理论,计算并通过实验验证了胆红素在524nm存在基准波长,进一步研究了此波长用于光谱修正的效果。此波长处的吸光度有望用于去除实际测量中的背景噪声和环境干扰,进而提高胆红素检测的精度。此外,研究了血红蛋白在胆红素检测波段的光学特性,分析了血红蛋白对胆红素光学检测尤其是在偏最小二乘回归(Partial Least Squares Regression,PLS )建模中对胆红素预测模型的影响。通过实验,发现胆红素水溶液中加入血红蛋白,使PLS建模过程中引进了血红蛋白的吸收噪声,增加了建模的主成分数量,改变了对建模贡献最大波长的位置;结合PLS评价指标发现血红蛋白的加入使模型预测均方根误差(Root Mean Square Error of Prediction,RMSEP)增大了24.90%,相关系数减小了0.29%,即血红蛋白降低了模型对胆红素浓度的预测能力和拟合能力。因此,实际测量中必须考虑去除血红蛋白的影响,才能进一步提高临床胆红素检测的精度,使胆红素检测适用于更广泛的人群。
[Abstract]:Abnormal metabolism of bilirubin can cause a series of diseases and seriously harm people's health. Therefore, serum bilirubin concentration is regarded as daily liver in clinic. It is an important physiological index in the examination of biliary function. At present, the main method of detection of bilirubin concentration in clinical human body is the invasive detection of serum bilirubin, which needs to take blood. Can cause pain discomfort and risk of infection, and can not be real-time monitoring of bilirubin concentration. Non-invasive optical bilirubin detection method with its non-invasive, rapid. Simple features play an important role in the detection of bilirubin in clinic, but the general accuracy of non-invasive optical bilirubin detection products such as percutaneous bilirubin monitor is not high. It can only be used for the primary screening of neonatal jaundice disease, which hinders the promotion and use of the product. In order to improve the accuracy of bilirubin detection using floating benchmark theory, the characteristics of noninvasive optical bilirubin detection were studied. The optical parameters of various skin models related to bilirubin detection were calculated and summarized. Monte Carlo simulation was carried out at a specific detection wavelength in the visible light band, and the distribution of diffusive reflected light at different radial positions with the change of bilirubin concentration was obtained, and the reference position of bilirubin detection could not be found. This is due to the unique optical properties of bilirubin in the visible light band. Based on the theory of reference wavelength in pure absorption medium, the existence of reference wavelength of bilirubin at 524nm was calculated and verified by experiments. The absorbance at this wavelength is expected to be used to remove background noise and environmental interference in actual measurement and improve the accuracy of bilirubin detection. In addition, the optical properties of hemoglobin in bilirubin detection band were studied. The optical detection of bilirubin by hemoglobin was analyzed, especially in partial Least Squares Regression. The effect of PLS on bilirubin prediction model was studied. It was found that hemoglobin was added to bilirubin solution, which led to the introduction of hemoglobin absorption noise in the modeling process of PLS. The number of principal components in modeling is increased, and the position of maximum wavelength for modeling is changed. Combined with the PLS evaluation index, it was found that the root Mean Square Error of Prediction could be predicted by adding hemoglobin. RMSEPs increased 24.90 percent, and the correlation coefficient decreased 0.29, that is, hemoglobin reduced the model's ability to predict and fit bilirubin concentrations. In order to improve the accuracy of clinical bilirubin detection, the effect of hemoglobin removal must be considered in the actual measurement, and the bilirubin detection can be applied to a wider range of people.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318.51

【参考文献】