皮肤组织癌变特性的太赫兹光谱研究

发布时间：2018-05-23 19:18

本文选题：太赫兹时域光谱(THz-TDS) + 皮肤癌变组织　；参考：《天津大学》2012年硕士论文

【摘要】：太赫兹时域光谱(THz-TDS)技术是一种新型的远红外相干光谱技术。近年来,太赫兹波科学与技术越来越多的引起了人们的关注。21世纪的生物与医学研究,成为人类科学技术研究中最为活跃的领域之一。太赫兹波具有诸多优越性,生物大分子振动及转动能级对应的特征谱位于太赫兹波段,因此太赫兹波在生物组织研究中得到了广泛的应用,在癌变组织的无损诊断方面有着巨大的应用潜力。本文分别介绍了THz波谱的特点,以及太赫兹波光谱技术对DNA的无标记探测研究、对氨基酸等生物大分子的振动波谱研究等的进展,并指出了太赫兹波谱技术在生物领域研究应用中有待解决的问题。特别重点阐述“水”成分对太赫兹波谱的影响,并对“水”与太赫兹波的作用机理进行了较为深入的分析。同时,通过实验验证了水含量的不同对太赫兹的吸收光谱带来的差异。基于人体健康组织与病变组织在水分以及其他组织成分的不同带来的对太赫兹波谱的吸收差异性,开展了太赫兹光谱和光谱成像技术研究人体皮肤组织的癌变分辨的工作。首先,利用透射式THz-TDS技术对石蜡封装以及新鲜切片状态的人体皮肤正常及癌变组织,进行了光谱探测实验研究,直接获得透过样品的太赫兹时域光谱,光谱携带了样品与太赫兹脉冲相互作用的电场强度和位相信息。通过对样品不同探测点的时域波谱进行对比分析发现,时域中幅值能够较好的反映健康组织与癌变组织的差异。然后对时域谱线作傅里叶变换(FFT)得到样品的频域谱,并对反映样品性质的太赫兹光学参数吸收系数及折射率等进行推导计算。对于不同的病变样本,区分癌变区域与非癌变区域的频率段表现出一定的差异性,在石蜡封装的1号样品在1.4THz频段的吸收系数能够将病变组织与健康组织进行区分,2号样品在1.5THz处的表现出较为明显的差异。经过对样品的规律性的多点扫描,选取不同频段中的参数对进行太赫兹光谱成像,并与病理学的诊断结果进行比对,结果表明太赫兹时域光谱成像在一定程度上合理反映了样品的病变情况。为进一步对频谱数据进行挖掘,本文结合模式识别,运用主成分分析法对光谱信息进行降维处理,同时用偏最小二乘法对癌变组织与健康组织进行建模与定性区分,在1.4THz-1.7THz处达到良好的建模分类效果,相关系数达0.983148。
[Abstract]:Terahertz time domain spectroscopy (THz-TDSs) is a new far-infrared coherent spectrum technique. In recent years, terahertz wave science and technology have attracted more and more attention. The biological and medical research in the 21st century has become one of the most active fields in human science and technology research. Terahertz wave has many advantages. The characteristic spectrum of vibrational and rotational energy levels of biological macromolecules is located in the terahertz band, so terahertz wave has been widely used in the study of biological tissue. It has great potential in nondestructive diagnosis of cancerous tissue. In this paper, the characteristics of THz spectrum and the progress of unlabeled detection of DNA by terahertz spectroscopy and vibrational spectrum of biological macromolecules such as amino acids are introduced. The problems to be solved in the research and application of terahertz spectroscopy in biological field are pointed out. The influence of "water" on terahertz spectrum is discussed, and the mechanism of "water" and terahertz wave is analyzed. At the same time, the differences of the absorption spectra of terahertz with different water content were verified by experiments. Based on the difference in the absorption of terahertz spectrum between healthy tissues and diseased tissues in water and other tissue components, terahertz spectroscopy and spectral imaging techniques were carried out to study the canceration resolution of human skin tissues. Firstly, the spectral detection of normal and cancerous tissues of human skin in paraffin encapsulation and fresh slice state was carried out by transmission THz-TDS technique. The THz time-domain spectra of the samples were obtained directly. The spectrum contains the field intensity and phase information of the interaction between the sample and the terahertz pulse. By comparing and analyzing the time-domain spectra of samples at different detection points, it is found that the amplitudes in time domain can reflect the difference between healthy tissues and cancerous tissues. The frequency domain spectra of the samples are obtained by Fourier transform Fourier transform (FFTFT), and the absorption coefficient and refractive index of terahertz optical parameters which reflect the properties of the samples are deduced and calculated. For different pathological samples, there are some differences in the frequency segments between cancerous regions and non-cancerous regions. The absorption coefficients of the paraffin encapsulated sample 1 in the 1.4THz band can distinguish the diseased tissue from the healthy tissue, and sample 2 shows obvious difference in 1.5THz. After the regular multipoint scanning of the sample, the parameters in different frequency bands were selected for terahertz spectral imaging, and the results were compared with the pathological diagnostic results. The results show that terahertz time domain imaging can reflect the pathological changes of the samples to some extent. In order to further mine spectrum data, this paper combines pattern recognition, uses principal component analysis method to reduce the dimension of spectral information, and uses partial least square method to model and qualitatively distinguish cancerous tissue from healthy tissue. Good modeling and classification effect was achieved at 1.4THz-1.7THz, and the correlation coefficient was 0.983148.
【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】：TH773;R739.5

【参考文献】