基于分布式近红外多波长传感器信息融合的无创血糖监测系统
发布时间：2019-04-03 20:36
【摘要】：人体血糖浓度的有效控制对一个人的健康至关重要,这是因为血糖是人体能量的直接来源,不正常的血糖浓度可能导致一系列的医学综合症,给病人带来身体上的损伤。当今医院常用的生化血糖检测系统能够提供比较准确的血糖浓度测量值,但这种方法的最大缺陷是需要进行血样采集,不能进行连续与在线监测。这种有创的血糖检测方法不仅给病人带来身体上的疼痛,而且可能传染一些体液传染性疾病。因此,无创的血糖监测系统成为研究的热点。 本文提出了一种新的无创血糖监测方法—基于分布式近红外多波长传感器阵列及信息融合技术的无创血糖监测。本论文的特色在于把多波长激光阵列及多传感器信息融合技术应用于血糖浓度监测中。这种人体无创红外多波长传感器阵列信息融合的监测方法能弥补单波长检测的缺陷,使其能从多种干扰因素中动态得到足够的可用血糖信息。同时该系统,采用了分布式监测部位设计的思想,把采集点设置在不同的部位,能避免多组波长在同一部位之间相互干扰,从而影响血糖连续浓度监测的准确性。在系统中采用了多波长传感器信息融合的思想,用BP神经网络的方法融合多波长传感器阵列采集的信息来建立标定与测量模型。为了实现连续监测的目的,针对红外多波长传感器之间的各种组合工作方式,分别建立标定模型,这样做的目的是实现多窗口血糖吸收峰值检测信息的融合,避免血糖与其它物质的重叠吸收所造成的信息误差；也可以在工作方式上实现单组传感器组合轮换扫描工作,可以有效避免全部激光器件长时间工作发热损坏的风险。 本论文的实验工作主要分为葡萄糖溶液模拟检测的仪器模型实验和人体血糖无创检测与连续监测实验。在这两组实验中分别得到了1200组和142组实验数据,并将其样本用于建立标定模型。在人体无创血糖监测实验中,建立的标定模型的均方根预测误差(RMSEP)为0.088mmol/L,与商品家用血糖仪检测数据的相关系数(CC)为0.94.,与单波长血糖浓度检测方法相比,本文设计的方法能很好地提高血糖检测准确性和精度,并可用于连续监测。因此,该论文对于血糖无创监测研究及改善糖尿病人的自我治疗与血糖控制有重要的实用价值和科学意义。
[Abstract]:The effective control of human blood glucose concentration is very important to one's health because blood sugar is the direct source of human energy. Abnormal blood glucose concentration may lead to a series of medical syndrome and bring physical damage to patients. The biochemical blood glucose detection system commonly used in hospitals today can provide more accurate blood glucose concentration measurement, but the biggest defect of this method is that blood samples need to be collected and can not be continuously and on-line monitoring. This invasive method of blood sugar detection not only brings physical pain to patients, but also can infect some humoral infectious diseases. Therefore, the non-invasive blood glucose monitoring system has become the focus of research. In this paper, a new noninvasive blood glucose monitoring method based on distributed near infrared multi-wavelength sensor array and information fusion technology is proposed. The characteristic of this paper is that multi-wavelength laser array and multi-sensor information fusion technology are applied to blood glucose concentration monitoring. This non-invasive monitoring method of infrared multi-wavelength sensor array information fusion can make up for the defect of single-wavelength detection and can dynamically obtain enough available blood glucose information from a variety of interference factors. At the same time, this system adopts the idea of distributed monitoring part design, and sets the collection points in different parts, which can avoid the interference between multiple wavelengths in the same part, and thus affect the accuracy of continuous monitoring of blood glucose concentration. The idea of multi-wavelength sensor information fusion is adopted in the system, and the calibration and measurement model is established by using the method of BP neural network to fuse the information collected by the multi-wavelength sensor array. In order to achieve the purpose of continuous monitoring, the calibration models are established for various combined working modes of infrared multi-wavelength sensors. The purpose of this calibration model is to achieve the fusion of multi-window blood sugar absorption peak detection information. Avoid the information error caused by the overlapping absorption of blood sugar and other substances; It can also work in a single group of sensors combined with rotating scanning, which can effectively avoid the risk of thermal damage caused by long-time operation of all laser devices. The experimental work of this thesis is mainly divided into the instrument model experiment of glucose solution simulation and the non-invasive and continuous monitoring experiment of human blood glucose. In these two groups of experiments, 1200 groups of experimental data and 142 groups of experimental data were obtained, and their samples were used to establish the calibration model. In the non-invasive blood glucose monitoring experiment, the root mean square prediction error (RMSEP) of the calibration model is 0.088 mmol / L, and the correlation coefficient (CC) between the calibration model and the measured data is 0.94.The correlation coefficient of the calibration model is 0.088 mmol / L. Compared with the single wavelength glucose detection method, the method designed in this paper can improve the accuracy and accuracy of blood glucose detection, and can be used for continuous monitoring. Therefore, this paper has important practical value and scientific significance for noninvasive monitoring of blood glucose and improvement of self-treatment and blood glucose control in diabetic patients.
【学位授予单位】：兰州大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318.6
[Abstract]:The effective control of human blood glucose concentration is very important to one's health because blood sugar is the direct source of human energy. Abnormal blood glucose concentration may lead to a series of medical syndrome and bring physical damage to patients. The biochemical blood glucose detection system commonly used in hospitals today can provide more accurate blood glucose concentration measurement, but the biggest defect of this method is that blood samples need to be collected and can not be continuously and on-line monitoring. This invasive method of blood sugar detection not only brings physical pain to patients, but also can infect some humoral infectious diseases. Therefore, the non-invasive blood glucose monitoring system has become the focus of research. In this paper, a new noninvasive blood glucose monitoring method based on distributed near infrared multi-wavelength sensor array and information fusion technology is proposed. The characteristic of this paper is that multi-wavelength laser array and multi-sensor information fusion technology are applied to blood glucose concentration monitoring. This non-invasive monitoring method of infrared multi-wavelength sensor array information fusion can make up for the defect of single-wavelength detection and can dynamically obtain enough available blood glucose information from a variety of interference factors. At the same time, this system adopts the idea of distributed monitoring part design, and sets the collection points in different parts, which can avoid the interference between multiple wavelengths in the same part, and thus affect the accuracy of continuous monitoring of blood glucose concentration. The idea of multi-wavelength sensor information fusion is adopted in the system, and the calibration and measurement model is established by using the method of BP neural network to fuse the information collected by the multi-wavelength sensor array. In order to achieve the purpose of continuous monitoring, the calibration models are established for various combined working modes of infrared multi-wavelength sensors. The purpose of this calibration model is to achieve the fusion of multi-window blood sugar absorption peak detection information. Avoid the information error caused by the overlapping absorption of blood sugar and other substances; It can also work in a single group of sensors combined with rotating scanning, which can effectively avoid the risk of thermal damage caused by long-time operation of all laser devices. The experimental work of this thesis is mainly divided into the instrument model experiment of glucose solution simulation and the non-invasive and continuous monitoring experiment of human blood glucose. In these two groups of experiments, 1200 groups of experimental data and 142 groups of experimental data were obtained, and their samples were used to establish the calibration model. In the non-invasive blood glucose monitoring experiment, the root mean square prediction error (RMSEP) of the calibration model is 0.088 mmol / L, and the correlation coefficient (CC) between the calibration model and the measured data is 0.94.The correlation coefficient of the calibration model is 0.088 mmol / L. Compared with the single wavelength glucose detection method, the method designed in this paper can improve the accuracy and accuracy of blood glucose detection, and can be used for continuous monitoring. Therefore, this paper has important practical value and scientific significance for noninvasive monitoring of blood glucose and improvement of self-treatment and blood glucose control in diabetic patients.
【学位授予单位】：兰州大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318.6
【参考文献】
相关期刊论文 前10条
1 郑怡磊;李远莉;刘欣;刘正英;杨伟;李尉;杨鸣波;;聚碳酸酯/羟基磷灰石纳米复合材料的力学性能与热性能[J];高分子材料科学与工程;2008年04期
2 丁东,张洪艳,王弼陡,唐玉国,申铉国;血浆中葡萄糖浓度的近红外光谱测量和分析[J];光学技术;2003年06期
3 王炜,卞正中,张大龙;红外无创血糖检测与阵列信号处理[J];红外技术;2003年02期
4 罗云瀚;徐可欣;;近红外无创伤血糖测量的组织光学基础研究[J];激光生物学报;2008年01期
5 王龙;肖君;骆清铭;龚辉;;基于近红外三波长检测葡萄糖水溶液浓度的两种预测模型比较[J];生物医学工程研究;2006年02期
6 李庆波,徐可欣,汪f，
本文编号：2453556
本文链接：https://www.wllwen.com/yixuelunwen/swyx/2453556.html
