光谱技术应用于乳腺癌诊断的研究

【作者】 姚淑霞；

【导师】 赵元黎；

【作者基本信息】 郑州大学 ， 测试计量技术及仪器， 2004， 硕士

【摘要】 光谱技术的发展，为我们能够较精确的分析和了解物质结构、成分及其含量提供了行之有效的方法。激光的出现，激光技术和光检测技术的发展，产生了高分辨率和高灵敏度的激光光谱技术，从而能使光谱技术在生物医学领域得到广泛的应用，成为研究生物分子、细胞、生命过程及医疗诊断和监测的重要手段。 二十多年来，人们用不同的激发波长，通过光谱分析技术，对动物和人体内、外肿瘤组织的各种光谱进行了大量的研究和观察，初步发现了肿瘤组织与正常组织的不同的光谱特征，并对光谱差别的来源作了一些探索。然而，由于各研究工作使用的仪器不同，提取的特征参数不同，判据计算也各有不同，致使诸多结果难以进行比较，更缺乏统一的诊断标准；尤其是对肿瘤与正常组织光谱差异产生的机制，现有的实验结果并没有在这个方面给出令人满意的关于病变组织分子水平上有明显特异性光谱来源的证据。 本论文主要通过研究乳腺癌患者医学样品的光谱和正常人的医学样品的光谱，比较他们之间的差异，运用光谱技术对乳腺癌的诊断和早期诊断提供依据。论文内容主要包括两大部分： 第一部分对癌症患者和正常人光谱的初步研究。 1、综述了光谱技术在医学诊断上的应用及其发展现状，对吸收光谱和拉曼光谱的基本原理进行了简要阐述。 2、探讨了一些实验因素对拉曼光谱的影响，发现用632.8nm光源激发血清光谱时需要选用低的透过率才能得到重现性好的光谱，且血清里不能加抗凝素。抗凝素的加入会改变血清光谱的特性。 3、通过几种激发光源的比较发现，绿光能激发出乳腺癌患者血清的拉曼光谱，而红光和蓝光只有荧光峰。 第二部分应用模式识别技术对光谱数据进行处理。 1、简要阐述了模式识别系统的基本原理及方法。 2、用根据相似性阈值和最小距离原则的简单聚类方法对乳腺癌和正常人血清的吸收谱进行分类识别，获得较好的分类效果。 3、分别利用最大最小距离聚类法和谱系聚类对乳腺癌患者和增生患者病理片的光谱进行聚类，并对两种方法设计的分类器加以比较，谱系聚类法的分类效果要明显好于最大最小距离聚类，达到了一定的识别率。更多还原

【Abstract】 The development in spectra technique provides an effective method to analyze and understand the structure , composition and content of the matter. With the appearance of laser and the development in the laser technique and inspective technique with laser, the resolution and sensitivity of spectrum technical were enhanced, so the spectrum technique can be widely used in biomedicine and has become an important method to study molecule , cell, life process and iatrical diagnosis and inspection .In the recent twenty years, scholar have been researching and analyzing kinds of spectra of tumor tissue. Researchers have find the different spectra character between tumor tissue and normal tissue and have done some research about the difference. Because of the difference in apparatus, characteristic parameter and estimation between different studies, many result are difficult compared, let alone produce unifiable diagnosis standard, especially about mechanism of tumour tissue different from normal tissue. Existing experimental result don’t offer satisfying evidences about weather the molecule in pathological change tissue has overt origin of special spectra or not.This paper study the spectra of medicine samples which come from breast cancer patients and normal people, compare with them between difference, and provide basis for early diagnosing breast cancer. This paper mainly includes two major parts:The first part is the preliminary study on cancer patients and normal people’s spectra.1. To summarize the use and development of spectra technique in medical diagnosis and elaborate the basic principle of absorption spectra Raman spectra.2. To explore some experimental factors affect Raman spectra and find it only use low penetration rate and antibiotic don’t be add up to sera, spectra with good reappearance ability can be acquired when light source stimulates sera spectra. Heparin can affect the character of sera spectra.3. Through comparing several light source, we find when use green light stimulate sera which come from breast cancer patients we can acquire Raman spectra but when use red or blue light we only acquire fluorescence peak.The second part handle spectra date with the system of pattern recognition.1. To elaborate the basic principle and method of system of pattern recognition.2. According to similar threshold value and minimum range principle, we classify and distinguish the absorption spectra of breast cancer patients and normal people’s sera and acquire better-classified effect.3. We respectively use minimax distance clustering and hierarchical clustering to cluster the spectra of breast cancer patients and hyperplasia patients’ pathology piece and compare two sorters .the effort of hierarchical clustering method is much better than that of minimax distance clustering.更多还原