舌色分类的量化研究

【作者】 杨新宇；

【导师】 梁嵘；

【作者基本信息】 北京中医药大学 ， 中医诊断学， 2012， 博士

【摘要】 目的本研究拟采用色度学方法,建立基于人眼视知觉特性的、反映舌色色貌的分类体系,并提取典型舌色色样,通过舌色分类研究,提高数字舌诊的客观化与规范化程度。方法舌色分析是通过量化舌的颜色特征,准确地表达和传递舌的颜色信息,为临床诊断提供依据。舌色的客观化度量问题最终是颜色的定量度量,现代颜色科学为其提供了理论支持与技术手段。1.基于聚类分析的舌色分类研究,根据文献调研估算样本量为1500-2500例,同时满足平均每种舌象不少于30例。运用色彩管理技术对数字舌图进行色彩还原,然后选择CIElab、lch均匀颜色空间描述舌图颜色信息,包括色相、明度、饱和度,建立数据库。应用聚类分析进行舌色分类研究。2.基于CIElab、lch均匀颜色空间开展舌色色样研究,根据舌色的颜色信息,确定完整覆盖舌色的颜色分布区域,等距划分颜色空间,提取基于人眼视觉特征的舌色标准色样。3.平和体质舌色分类特征调查,将舌色标准色样与平和质人群舌色色度空间的比较,观察两者色度值差异,同时根据色差判断两组舌色的异同。4.舌色分类研究成果的验证性研究采用数字化评价方法,比较舌色分类数据与临床采集样本人群色度空间、聚类中心色度参数。结果1.运用色度学的研究结果表明：以淡红舌为比较的基准,舌色色度数据的变化可以体现出视觉判断下舌色表现出的连续、渐变的特征。2.基于视觉判断的舌色分类研究发现,舌色聚集于红黄色域空间内,男性舌色偏暗,女性舌色偏于浅淡。3.舌色色度学数据整体收敛在相对集中的区域内部,根据舌色拾取软件的命名,以粉色为主要聚集中心。其次,数据簇集为约11大类,彼此间分割清晰,类内聚集状态均匀。4.本研究发现非人工判读分类结果与人工判断结果存在不一致的地方,淡白舌分类两者比较,以人工判断为基准,吻合率达到55.56%,其它舌色吻合度情况如下：淡红舌31.08%、红舌30.75%、暗红舌23.99%、绛舌47.37%、淡暗舌19.17%、紫暗舌37.97%。5.平和体质人群舌色分布总体特征与体检人群相接近,不同舌色数据在颜色空间中呈现出特征性聚集状态,数据簇集趋势明显,不同数据集合之间边界相对清晰,整体分布集中,边缘位置散在数据量小,空间分布图显示两组数据类聚集特征相同。6.临床采集样本舌色色域范围较大,标准研究划定的舌色范围,尚未完全涵盖临床所见舌色,仍需进一步扩大研究范围,广泛采集舌色样本。结论1.以人眼视觉判断为基础,应用聚类分析方法进行舌色分类研究,获得舌色分类数据的一般特点,并获取典型舌色的样品数据。从而揭示能够完整和等距覆盖舌色空间的颜色规律,为提取符合人眼颜色视觉特性的舌色色样奠定基础。2.将色度学的方法用于分析基于人眼视觉的舌诊,可以借助色度值描述舌诊信息,为舌诊经验的理论提升和客观分析提供了共同平台；应用数学方法获取典型色的色度值,为进一步研究提供了方法学基础。3.设定标准条件进行舌色的采集与分析,对于舌色分类研究的准确性和可重复性具有决定作用。更多还原

【Abstract】 PurposeIn this study, to make improvement at the objective of the digital tongue inspection and degree of standardization, the colorimetry method was adopted to establish classification system that can reflect the color appearance of the tongue color based on the human eye visual perception characteristics, and extract the typical tongue color samples.MethodTongue color analysis is to quantify the color characteristics of thetongue, accurately express and pass the tongue’s color information, provide the basis for clinical diagnosis. The objective measure of the tongue color is ultimately a quantitative measure of the color, the modern color science provides a theoretical support andtechnical means.1. Tongue color classification based on cluster analysis According to the literature survey, the sample size is estimated to be 1500-2500 cases, and the number of each tongue color samples at least 30 cases. Color correction was carried out for each digital image, select CIE lab, CIE lch uniform color space and munsell sequential color system to describes the color information of the tongue Fig. A database was created and the data included hue, lightness, saturation and Munsell number, the application of cluster analysis was used in the study of the tongue color classification.2. Studies of standard Tongue color samples According to the color information of the tongue, we determined the complete coverage of the tongue color distribution area, equidistant division of the color space, extraction the tongue color standard samples based on human visual characteristics.3. The survey of tongue color classification characteristics in mild constitution We compared tongue color standard samples and mild constitution crowd’s, observed chromaticity value of the difference between them, determined the similarities and differences of the two sets of tongue color according to color difference.4. Clinical validations of the tongue color classification research achievements Digital evaluation method was used to verify the accuracy of the classification criteria of the tongue color, and the comparison of the color space and cluster centers were carried out between the tongue color disaggregated data and clinical data.Results 1. By using of colorimetry this study found tongue color change in a continuous gradient range concentrate the pink color area.2. The study of tongue color classification based on visual judgment foundthat the tongue color gathered in the space of the red and yellow fields. Male tongue color is dim, female tongue Hue in pale.3. As a whole, tongue color chromaticity data is in a relatively concentrated area. According to the name given by the color Picker software, the pink was the main gathering center of the tongue color. Secondly, the data clustered as about 11 categories, split between each other clearly, aggregation states uniformity within a class.4. This study found that not artificial interpretations of classification results are inconsistent with the artificial judge outcome. The consistent rate of Pale tongue classification between the two groups was 55.56%. other tongue color consistent as follows:light red tongue 31.08%, red tongue 30.75%, dark red tongue 23.99%, purple tongue 47.37%, short dark tongue 19.17%, dark purple tongue 37.97%.5. The overall characteristics of tongue color distribution in mild constitution crowd were closed to medical crowd. Different tongue color data in the color space showing a characteristic of the aggregation state, the central tendency of the data is significant, the boundary between the different data sets was relatively clear, and there is only small amounts of data scattered in the edge position. The spatial distribution of the figure of two sets data class showed the same characteristics.6. The tongue color gamut of clinical collect samples was larger than that of medical populations. The standard tongue color range not yet fully covered the clinical findings, and it need to be further expand the scope of the study and widely collected samples of tongue color.Conclusion1. Application of cluster analysis method in the study of tongue color can reveal the law of tongue colorspace.2. Colorimetric method for the analysis of tongue diagnosis based on human visual can make use of the color values to describe the tongue diagnosis information, provide a common platform for the theory of tongue diagnosis experience and objective analysis; the application of mathematical methods to obtain the color values of the typical color provides a methodological basis for further research.3. A standard condition for the collection and analysis of the tongue color play an important role in tongue color classification study.更多还原