提高动态光谱信噪比的方法及应用

发布时间：2018-04-29 07:21

本文选题：血液成分无创检测 + 动态光谱　；参考：《天津大学》2014年博士论文

【摘要】：血液成分的无创检测，不仅对各种疾病的诊断，糖尿病、贫血等慢性疾病的管理，围手术期或急诊患者的监护具有重要意义，还可以实现疾病的早期筛查，节省医疗资源，促进环保。研究无创血液成分检测可以推动无创生物医学信息传感、微弱信号检测、基础医学以及临床医学的发展。光谱技术以其过程便捷、无痛无创、以及原理上高速、高精度、信息多维化等优点，成为最具应用前景的检测手段。在体光谱检测，满足安全性、舒适性的基础上，不仅要面临着信号微弱、光谱重叠、漂移等因素的影响，还需要克服个体差异、测量条件、被测者心理状态变化等不确定因素的影响。因此，提高信噪比是实现无创血液成分检测的首要目标。动态光谱方法，基于透射的光电容积脉搏波，利用动脉充盈程度对光谱吸收的改变来直接提取多个波长下反映动脉血液成分的光密度，从理论上降低了个体差异和测量条件的影响，相比其他方法具有显著的优势。本文首先介绍了动态光谱的基本原理，在考虑组织特性的基础上通过蒙特卡罗仿真（Monte Carlo, MC）获取不同血红蛋白浓度下的动态光谱值，并计算与血红蛋白浓度的相关系数，从而验证了动态光谱方法的可行性；其次通过搭建的基于光栅光谱仪与溴钨灯的测量系统采集实际临床数据，通过偏最小二乘方法建立校正模型，实验结果进一步验证了动态光谱方法用于无创血液成分检测的可行性。但测量精度还有待提高，需要进一步提高信号检测与处理中的信噪比，论文主要从以下四个方面进行。从信号传感的角度提高信噪比。对于数据采集系统，通过MC仿真比较窄平圆光束、宽平圆光束以及宽光纤光束这三种光照条件下动态光谱的差异，进而分析由光照方式引入光路径差异对动态光谱的影响，在此基础上获取两种实际光源（溴钨灯与超连续激光器）下的动态光谱，验证了仿真结果，即系统中使用细光束更有利于提高光电容积脉搏波的灵敏度。对于数据采集方式，提出动态光谱优化采集方法“双采样”，与传统方法对比实验结果表明，，双采样方法增加了采集数据的动态范围，显著提高动态光谱的信噪比，拓宽了光谱采集的波段。从动态光谱提取的过程中提高信噪比。通过理论与实际提取效果，对比分析频域提取法与单沿提取法，结果验证了两种方法的一致性与可行性，以及单沿法在去除脉搏波片段干扰的优势。基于此，结合两种方法的优势，进一步提出一种基于快速数字锁相算法的提取方法，通过将信号分段利用快速的数字锁相算法实时提取片段动态光谱，在大幅度降低了计算量的同时，更有助于剔除无效数据，提高动态光谱的精度。仿真与实际测量实验，充分验证了方法的可行性，评估了该方法在提取速度与精度上的优势。从信号处理与建模的角度提高信噪比。对于散射的影响，通过三种模型的MC仿真分析其影响方式与程度。对于光谱分析，提出用于研究光谱分析内在机制的编码方法，并通过构造三种组分混合物质的光谱进行建模分析阐述方法的实现过程。结果表明，编码方法能够用来分析不同组分吸光度大小关系对光谱分析的影响规律与影响程度，有助于波段优选，准确地把握建模方法的内在原理。此外，分析水的吸收峰对血红蛋白含量检测的影响，实验结果表明建模时使用的波段覆盖主要的非测量成分吸收峰时能够提高建模的精度，为降低非目标组分的影响、优选波段提供了参考。设计了基于LED和DSP的便携式测量系统。系统采用光调制技术，利用结合过采样的快速数字锁相算法实现八通道光电容积脉搏波的同步高精度测量，简化了外部硬件电路，使得系统更加灵活，便于小型化与低成本化。文中对所采用的结合过采样的快速锁相算法进行了详细的理论分析，通过仿真与实测实验验证了算法的可行性，评估其在微弱信号检测中的性能。同时，将这种快速高精度算法推广到一般多通道传感测量中，并针对载波频率设置、邻道串扰等问题，提出了一般应用中的参数设置与实现方法，使得方法具有普适性。论文通过血红蛋白无创检测验证动态光谱可行性的基础上，从信号检测、提取、处理与建模分析，以及便携式检测装置的设计这四个方面提出了一系列的方法与措施提高光谱信号检测的精度与灵敏度，为动态光谱血液成分无创检测建立完整的方法体系，同时解决在体光谱检测、微弱信号检测中的一些共性问题，为其他相关应用提供借鉴与参考。
[Abstract]:Noninvasive detection of blood components is of great significance not only to the management of the diagnosis of various diseases, diabetes, anemia and other chronic diseases, but also for the monitoring of the perioperative period or emergency patients. It can also realize early screening of the disease, save medical resources and promote environmental protection. The development of weak signal detection, basic medicine and clinical medicine. Spectral technology has the advantages of convenient process, painless and noninvasive, high speed, high precision, multidimensional information and so on. It has become the most promising detection means. On the basis of body spectrum detection, it is not only faced with weak signal and spectral weight on the basis of safety and comfort. The influence of superposition, drift and other factors also needs to overcome the influence of individual differences, measurement conditions, and changes in the mental state of the subjects. Therefore, improving the signal to noise ratio is the primary goal for the realization of non-invasive blood components detection. To directly extract the light density that reflects the arterial blood components at multiple wavelengths, theoretically reduces the individual difference and the influence of the measurement conditions, and has a significant advantage compared with other methods.
In this paper, the basic principle of dynamic spectroscopy is introduced. The dynamic spectral values of different hemoglobin concentrations are obtained by Monte Carlo simulation (Monte Carlo, MC) on the basis of the characteristics of the tissue, and the correlation coefficient of hemoglobin concentration is calculated, thus the feasibility of the dynamic spectral method is verified. Secondly, the structure based on light is built. The measurement system of the grating spectrometer and the bromo tungsten lamp collects the actual clinical data, and establishes the correction model by the partial least square method. The experimental results further verify the feasibility of the dynamic spectral method for non-invasive blood component detection, but the measurement precision still needs to be improved and the signal to noise ratio in the signal detection and processing needs to be further improved. It should be carried out in the following four aspects.
From the angle of signal sensing, the signal to noise ratio is improved. For data acquisition system, MC simulation is used to compare the difference of dynamic spectrum between narrow flat circular beam, wide flat circular beam and wide optical fiber beam under the three illumination conditions, and then analyze the influence of light path difference on the dynamic spectrum by illumination mode. On this basis, two kinds of actual light sources (bromine) are obtained. The dynamic spectra under the tungsten lamp and the supercontinuous laser verify the simulation results, that is, the use of fine light beams in the system is more conducive to improving the sensitivity of the photoelectric volume pulse wave. For data acquisition, a dynamic spectral optimization acquisition method "double sampling" is proposed. The double sampling method increases the number of sampling numbers with the traditional method. According to the dynamic range, the signal-to-noise ratio of the dynamic spectrum is significantly improved, and the band of spectral acquisition is broadened.
In the process of dynamic spectral extraction, the signal to noise ratio is improved. The results of theoretical and practical extraction are compared and analyzed in frequency domain and single edge extraction. The results verify the consistency and feasibility of the two methods, as well as the advantages of the single edge method in removing the interference of the Mai Bobo fragment. Based on this, we combine the advantages of the two methods to further propose a basis. The method of extracting the fast digital phase locked algorithm, by using the fast digital phase locked algorithm to extract the dynamic spectrum in real time by using the fast digital phase locked algorithm, is more helpful to eliminate the invalid data and improve the accuracy of the dynamic spectrum at the same time. The simulation and actual measurement experiments have fully verified the feasibility of the method and evaluated the feasibility of the method. The advantage of this method is to extract speed and precision.
From the point of view of signal processing and modeling, the influence of signal to noise is improved. The influence mode and degree of the three models are analyzed by the simulation of the three models. For spectral analysis, the coding method used to study the internal mechanism of spectral analysis is put forward, and the realization of the method is realized by constructing the spectrum of the three components of the mixture. The results show that the coding method can be used to analyze the influence of different components of absorbance on spectral analysis and the degree of influence. It is helpful to optimize the wave band and grasp the intrinsic principle of the modeling method. In addition, the influence of the absorption peak of water on the detection of hemoglobin content is analyzed. The experimental results show the wave band used in modeling. The accuracy of modeling can be improved by covering the main absorption peaks of non measured components, providing a reference for reducing the influence of non target components and optimizing wavebands.
A portable measurement system based on LED and DSP is designed. The system uses the optical modulation technology and the fast digital phase locked algorithm combined with over sampling to realize the synchronous high precision measurement of the eight channel optical capacitive pulse wave. It simplifies the external hardware circuit, makes the system more flexible, is convenient for miniaturization and low cost. The fast phase locked algorithm for over sampling is analyzed in detail. The feasibility of the algorithm is verified by the simulation and the measured experiment, and the performance of the algorithm is evaluated in the weak signal detection. At the same time, this fast and high precision algorithm is extended to the general multi-channel sensing measurement, and the problem of carrier frequency setting, adjacent channel crosstalk and so on is proposed. The method of parameter setting and implementation in general application makes the method universal.
The paper proposes a series of methods and measures to improve the accuracy and sensitivity of spectral signal detection on the basis of the feasibility of non invasive detection of hemoglobin to verify the feasibility of dynamic spectrum, from four aspects: signal detection, extraction, processing and modeling analysis, and the design of portable detection devices, which are established for noninvasive detection of dynamic spectral blood components. The complete method system also solves some common problems in in vivo spectrum detection and weak signal detection, and provides reference for other related applications.

【学位授予单位】：天津大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：R318.04

【参考文献】