血液中的其他成分对近红外无创血糖检测影响的研究

发布时间：2018-06-26 02:18

本文选题：近红外光谱 + 无创血糖检测　；参考：《天津大学》2012年硕士论文

【摘要】：利用近红外光谱分析技术进行人体血糖浓度测量具有巨大的发展前景和重要的现实意义,但是该技术在实际应用中却遇到了相当的难度,其测量精度极易受到其它因素的影响。其中一个重要因素是血液中的其他成分会对血糖浓度的近红外无创检测产生影响。基于此,论文首先从描述血液中其他成分与血糖的关系出发,对人体血液成分进行了生化分析、临床分析及光谱分析。结果表明:正常人和糖耐量受损患者血液中葡萄糖与其他各成分的变化关系不显著;而糖尿病患者血液中葡萄糖与胆固醇和总蛋白的浓度变化却呈现负相关,尽管与其他成分的变化关系不显著血液中其他成分与葡萄糖的近红外吸收光谱有重叠,其中胆固醇和葡萄糖的近红外吸收光谱重叠最严重。其次,以胆固醇为研究对象,通过实验研究胆固醇对近红外无创血糖检测的影响。实验方法为:首先,测量不同浓度的葡萄糖溶液和胆固醇溶液的近红外吸收光谱,比较二者光谱谱峰的差异;然后,在不同浓度葡萄糖溶液中加入一定浓度的胆固醇,观察溶液近红外光谱的变化情况,并求出谱峰位置处的吸光度变化率;最后,采用偏最小二乘(PLS)法对加入胆固醇前后的葡萄糖溶液光谱数据进行处理,建立葡萄糖浓度的预测模型,并对各模型进行分析评价。实验结果表明,胆固醇对血糖浓度的近红外检测结果有影响,具体的影响关系为葡萄糖的测量浓度随着胆固醇的浓度增加而线性增大,并给出了胆固醇影响的定量关系式。最后,基于以上研究,给出了基于基准波长来修正胆固醇对血糖近红外检测影响的修正方法:首先用基准波长法精确地测量混合溶液中胆固醇的浓度,然后求出由胆固醇引起的葡萄糖浓度测量结果的增量,最后从葡萄糖浓度的测量值中扣除此增量。并采用此方法对近红外光谱下血糖浓度的预测效果进行了实验研究。实验结果表明,用此方法修正后葡萄糖预测模型的预测精度有了一定的提高,提高幅度最高可达38.36%,平均为29.42%。初步证明了本文提出的修正方法具有可行性和有效性。
[Abstract]:The application of near infrared spectroscopy (NIR) to the measurement of human blood glucose concentration has great development prospects and important practical significance. However, it has encountered considerable difficulties in practical application, and its measuring accuracy is easily affected by other factors. An important factor is the effect of other components in the blood on near-infrared non-invasive detection of blood glucose levels. Based on this, the paper first describes the relationship between other blood components and blood sugar, and carries out biochemical analysis, clinical analysis and spectral analysis of human blood components. The results showed that there was no significant relationship between blood glucose and other components in normal subjects and patients with impaired glucose tolerance, but there was a negative correlation between the concentration of glucose and cholesterol and total protein in diabetic patients. Although there is no significant relationship between the changes of other components and the other components of blood, there is overlap between the other components of blood and the near infrared absorption spectra of glucose, among which cholesterol and glucose have the most serious overlap in near infrared absorption spectra. Secondly, the effect of cholesterol on non-invasive near-infrared blood glucose detection was studied experimentally. The experimental methods are as follows: first, the near infrared absorption spectra of different concentrations of glucose solution and cholesterol solution were measured, and the differences between the two spectral peaks were compared. The changes of near infrared spectra of the solution were observed, and the absorbance change rate at the peak position was calculated. Finally, the spectral data of glucose solution before and after cholesterol addition were processed by partial least squares (PLS) method. The prediction model of glucose concentration was established, and each model was analyzed and evaluated. The experimental results show that cholesterol has an effect on the results of near-infrared detection of blood glucose concentration. The specific relationship is that the measured concentration of glucose increases linearly with the increase of cholesterol concentration, and the quantitative relationship of the effect of cholesterol is given. Finally, based on the above research, a correction method based on the reference wavelength to correct the effect of cholesterol on the near infrared detection of blood glucose is given. Firstly, the concentration of cholesterol in mixed solution is accurately measured by the reference wavelength method. Then the increment of glucose concentration caused by cholesterol is obtained, and the increment is deducted from the measured value of glucose concentration. The prediction effect of blood glucose concentration in near-infrared spectrum was studied by using this method. The experimental results show that the prediction accuracy of the modified glucose prediction model has been improved to a certain extent, and the maximum range can be up to 38.36, with an average of 29.42. It is proved that the proposed method is feasible and effective.
【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】