【作者】 蔡永军；

【导师】 虞启琏；

【作者基本信息】 天津大学 ， 生物医学工程， 2004， 硕士

【摘要】 尿液分析作为临床检验的三大常规检查之一,以其标本收集方便,病人无痛苦,具有广泛的临床应用。然而现行的尿液分析方法主要借助于化学方法来进行分析,这种方法费时、费力、有污染,而且破坏标本。本文的目的就在于寻求一种快速的,不破坏样品,不添加任何试剂的纯绿色分析方法。本文分别采用中红外ATR分析技术,近红外透射技术以及近红外傅立叶变换拉曼光谱技术进行了尿液分析的研究探索,使用多元统计工具来进行尿液的定量分析。中红外光谱处于物质的基频振动区,因此信噪比和精度较高。对葡萄糖水溶液通过多变量回归分析,其预测精度RMSEP可以达到13.69mg/dl,相关性R接近1。但在尿液多成分体系中各种成分会相互影响,建模精度有所降低。近红外光谱的测试条件更加灵活,近红外的光谱信息也更加丰富。通过0.7mm和1.5 mm两个光程下分别使用4200～4800cm-1和5500～6500cm-1波段进行建模分析,验证了理论推导的最佳光程长理论。为了满足临床实时分析的要求,提高分析效率,本研究中优选了10个和6个建模参考波长,为下一步快速分析模型的建立奠定了基础。近红外傅立叶变换拉曼光谱可以很方便的得到尿液中各种成分的纯净物的拉曼散射光谱,但是由于系统使用1064nm(9399cm-1)的红外激发光,使得散射光进入水的近红外吸收区,产生了很强的非线性,因此不适合做含水体系的物质成分分析。本研究对于尿液分析所能达到的精度只能起病理过筛的作用,要达到临床实用还需要进行大量的试验,优化定量分析的模型。更多还原

【Abstract】 As a normal tool in clinical analysis, urinalysis has been widely used in hospitalbecause urine is easy to be collected without pain. The current chemical urinalysismethod is time consuming, contaminative and destructive and reagentless method inorder to facilitate the clinical application of urinalysis.In this thesis, middle infrared attenuated total reflection (MIR-ATR) spectrum, nearinfrared (NIR) transmission spectrum and near infrared flourier transform Raman(NIR-FT-Raman) spectrum were introduced into urinalysis. Multivariate statistics wasused to analyze spectra quantitatively.MIR spectrum is based on substance fundamental frequency oscillation, so SNR(signal to noise ratio) is higher than other two methods. The results from glucosewater solution indicated that RMSEP (root mean square error of prediction), anindicater for precision, was 13.69mg/dl, and correlation coefficient (R) was ulmostone. At the same time, the other components of urine disturb the predictive precisionin real urine model.NIR spectrum has more flexible measure conditions and much more information. Thetwo optical path lengths 0.7mm and 1.5 mm was used to model at 4200~4800cm-1and 5500~6500cm-1 respectively. The results were in accodance with the theory ofoptimum optical path length. In order to facilitate real time application in clinic, bythe use of optimum wavelengths, two sets of wavelengths, ten and six wavelengths,were used to improve the modeling.NIR-FT-Raman spectra of sigle components in urine are easily obtained. However,1064nm laser used in instrument made scattering light influence the region of NIRwater absorption. That induced the nonlinearity in spectra, therefore NIR-FT-Ramanis not suitable for the analysis of component in water solution.The results presented didn’t reach clinical availability, and the methods introduced canbe used to pathologic urine screening. The further study should be done in order tooptimize model and thus improve the clinical availability of reagentless spectroscopyin urinalysis.更多还原