Raman spectroscopic analysis differentiates between breast cancer cell lines

Breast cancer incident rates are increasing in women worldwide with the highest incidence rates reported in developing countries. Major breast cancer screening approaches like mammography, ultrasound, clinical breast examination (CBE) and magnetic resonance imaging (MRI) are currently used but have their own limitations. Optical spectroscopy has attained great attention from biomedical researchers in recent years due to its non‐invasive and non‐destructive detection approach. Chemometrics is one of the powerful tools used in spectroscopic research to enhance its sensitivity. Raman spectroscopy, a vibrational spectroscopic approach, has been used to explore the chemical fingerprints of different biological tissues including normal and malignant types. This approach was used to characterize and differentiate two breast cancer and one normal breast cell lines (MDA‐MB‐436, MCF‐7 and MCF‐10A) using dispersive Raman spectroscopy. Raman spectra of the cell lines have revealed that basic differences in the concentration of biochemical compounds such as lipids, nucleic acids and protein Raman peaks were found to differ in intensity, and principal component analysis (PCA) was able to identify variations that lead to accurate and reliable separation of the three cell lines. Linear discriminant analysis (LDA) model of three cell lines was predicted with 100% sensitivity and 91% specificity. We have shown that a combination of Raman spectroscopy and chemometrics are capable of differentiation between breast cancer cell lines. These variations may be useful in identifying new spectral markers to differentiate different subtypes of breast cancer although this needs confirmation in a larger panel of cell lines as well as clinical material. Copyright © 2015 John Wiley & Sons, Ltd.     

Biomolecular component analysis of cultured cell nucleoli by Raman microspectrometry

Distinct cellular domains, such as structure–function compartments of the cell nucleus and cytoplasmic organelles, are responsible for numerous macromolecular processes essential for cell functions. Spectroscopic analysis of specific cellular domains opens a way for noninvasive characterization of their molecular content and monitoring of their function. Confocal Raman spectroscopy was employed here for characterization of the complex molecular organization of major structure–function compartment of the cell nucleus, the nucleolus. The Raman spectra obtained in the nucleoli were processed by biomolecular component analysis (BCA). BCA was used to determine the contribution of each major type of macromolecules (proteins, DNA, RNA and lipids) to the complex molecular composition of nucleoli. A notable cell‐to‐cell variability in the macromolecular composition of nucleolus was found. At the same time, we observed a correlation between the concentrations of major types of biomolecules in this nuclear compartment. In particular, the averaged concentration of RNA increases along with increase in protein concentration, while an inverse dependence between the concentrations of RNA and DNA was found. Variations in the nucleolar concentrations of lipids were also noticed. Manifestations in spectral variations of proteins for individual nucleoli, shown by BCA, are discussed and interpreted. We also compared utility of BCA and principal component analysis for biomolecular studies and conclude that BCA is a more powerful and informative technique for studies of macromolecular composition and its variations in specific subcellular domains. Copyright © 2012 John Wiley & Sons, Ltd.     

Different classification algorithms and serum surface enhanced Raman spectroscopy for noninvasive discrimination of gastric diseases

In this study, surface enhanced Raman spectroscopy (SERS) was used to investigate the spectral characteristics of blood serum for the purpose of diagnosing stomach diseases. SERS spectral data was collected from patients with atrophic gastritis, both pre‐operation and post‐operation gastric cancer, and from healthy individuals. Visual differences in the SERS spectra were observed between the four groups which indicate corresponding biomolecule concentration changes in blood. To further investigate the diagnostic ability of human serum, the spectral data was analyzed with three chemometric processes. These three methods extracted features and classified from the spectral data. Principal component analysis (PCA) was first performed to reduce the dimensionality of the original spectral data. Then, the classification methods support vector machine (SVM), linear discriminant analysis (LDA) and classification and regression tree (CART) were used for the evaluation of diagnostic ability. Accuracies of 96.5%, 88.8% and 87.1% were obtained for PCA‐SVM, PCA‐LDA and PCA‐CART, respectively. Copyright © 2016 John Wiley & Sons, Ltd.     

Discrimination of NPC cell lines associated with malignant types using Raman spectroscopy

Nasopharyngeal carcinoma (NPC) is subdivided into four categories according to both the characterization of the NPC cell and their radiotherapy reaction. To discriminate these kinds of NPC, the representative cultured cell types (CNE1, CNE2, 5-8F, and 6-10B) are explored by single-cell micro-Raman spectroscopy (RS). The collected spectra are analyzed and compared by principal component analysis (PCA) and linear discriminate analysis (LDA). The algorithm is shown to have the lowest specificity and sensitivity of 89 and 92% for NPC division. These results demonstrate that it is a potentially clinical application for Raman spectroscopy to diagnosis NPC tissue types.     

Study of both fingerprint and high wavenumber Raman spectroscopy of pathological nasopharyngeal tissues

High wavenumber (HW) Raman spectroscopy has weaker fluorescence background compared with fingerprint (FP) region. This study aims to evaluate the discrimination feasibility of nasopharyngeal non‐cancerous and nasopharyngeal cancer (NPC) tissue with both FP and HW Raman spectroscopy. HW Raman spectra of nasopharyngeal tissue were obtained for the first time. Raman spectra were collected to differentiate nasopharyngeal non‐cancerous (n = 37) from NPC (n = 41) tissues in FP (800–1800cm⁻¹), HW (2700–3100cm⁻¹), and integrated FP/HW region. First, to assess the utility of this method, the averaged Raman spectral intensities and intensity ratios of corresponding Raman bands were analyzed in HW and FP regions, respectively. The results show that intensities as well as the ratios of specific Raman peaks might be helpful in distinguishing nasopharyngeal non‐cancerous from NPC tissue with the HW Raman spectroscopy, as with FP Raman reported before. The multivariate statistical method based on the combination of principal component analysis–liner discriminant analysis (PCA‐LDA), together with leave‐one‐patient‐out, cross‐validation diagnostic algorithm, was used for discriminating nasopharyngeal non‐cancerous from NPC tissue, generating sensitivities of 87.8%, 85.4%, and 95.1% and specificities of 86.5%, 91.9%, and 89.2%, respectively, with Raman spectroscopy in the FP, HW, and integrated FP/HW regions. The posterior probability of classification results and receiver operating characteristic curves were utilized to evaluate the discrimination of PCA‐LDA algorithm, verifying that HW Raman spectroscopy has a positive effect on the differentiation for the diagnosis of NPC tissue by integrated FP/HW Raman spectroscopy. What's more, the potential of Raman spectroscopy used for differentiating different pathology NPC tissues was also discussed. The results demonstrate that both FP and HW Raman spectroscopy have the potential for diagnosis and detection in early nasopharyngeal carcinoma, and HW Raman spectroscopy may improve the discrimination of NPC tissue compared with FP region alone, providing a promising diagnostic tool for the diagnosis of NPC tissue. Copyright © 2015 John Wiley & Sons, Ltd.     

线性辨别分析与人工神经网络用于血清光谱对于肝癌、肝硬化的诊断

本文探测了人体血清的自体荧光-拉曼光谱,采用多元统计方法中的主元分析法(PCA)对光谱进行分析,并利用线性辨别分析(LDA)作为诊断算法,与此同时,用人工神经网络进行交叉认证。PCA-LDA的灵敏度和特异性分别为88.00%和79.14%,PCA-ANN的为89.29%和94.74%。     

基于拉曼光谱和多元统计分析技术的结肠癌早期诊断

为了寻找结肠癌的病理发展规律在拉曼光谱属性上的体现,采用共聚焦显微拉曼光谱仪对60例离体的正常结肠组织、腺瘤性息肉和腺癌组织的近红外拉曼光谱进行对比检测,初步探讨了三类组织的拉曼光谱特征及其改变的规律。结果表明:三类组织拉曼光谱的差异明显存在于830,855,1 032,1 210,1 323,1 335,1 445,1 450,1 655 cm^-1处,腺瘤性息肉的光谱大体位于正常组织与腺癌组织之间。以组织病理诊断为金标准,主成分分析结合线性判别分析技术建立的诊断算法区分3类组织的灵敏度分别为96.9%、85.7%和97.3%,特异性分别为82.8%、90%和92.3%。因此,拉曼光谱有潜力在分子水平上区分正常结肠组织、腺瘤性息肉和腺癌组织,有望成为结肠癌早期诊断的一种有效方法。     

单个鼻咽癌细胞的拉曼光谱分析的研究

利用激光镊子拉曼光谱系统研究了鼻咽癌细胞株和正常人鼻咽部气道上皮细胞株的单个细胞的拉曼光谱,对于每个细胞在不同部位测3个点。结果显示：正常细胞和癌细胞的平均拉曼光谱有显著差异：正常的细胞光谱强度比癌细胞的明显要高;正常细胞的1304和1336 cm-1处峰的强度比值为1.05,癌细胞的为1.22。用PCA主成分分析和DFA判别分析分别对单个细胞的平均光谱和不同位置所取得的单独光谱进行分析,结果发现：PCA和DFA均可以把癌细胞和正常细胞正确区分,对于单独光谱,DFA的效果更好一些。同时还发现同一个细胞中不同的光谱位置对PCA和DFA的区分度影响不是很大;PCA和DFA的图中还表明癌细胞的均匀度要比正常细胞的差。以上的研究均表明：激光镊子拉曼光谱可以成为区别正常鼻咽细胞和鼻咽癌细胞的有效手段。     

Raman spectroscopy as a tool for detecting mitochondrial fitness

Raman spectroscopy allows the molecular chemical analysis of whole living cells by comparing them to known Raman signatures of specific vibrational bonds. In this work we used Raman spectroscopy to differentiate between wild type yeast cells and mutants characterized by increased or reduced mitochondrial fragmentation. To associate mitochondrial fragmentation with biochemical markers, we performed Linear Discriminant Analysis (LDA) of whole cell Raman spectra (~50–100 cells/spectrum). We show that the long‐lived, less fragmented mutants fall into a significantly distant cluster from the wild type and short‐lived, more fragmented mutants. Clustering depends on respiratory growth and coincides with that of membrane phospholipids and some respiratory chain components. Spectral clustering is supported by enzymatic activity measurements of OXPHOS Complexes. In addition, we find that NAD(P)H autofluorescence also correlates with mitochondrial fragmentation, representing another likely aging biomarker, besides phospholipids and OXPHOS components. In summary, we demonstrate that Raman spectroscopy has the potential to become a powerful tool for differentiating healthy from unhealthy aged tissues, as well as for the prognostic evaluation of mitochondrial function and fitness. © 2016 The Authors Journal of Raman Spectroscopy Published by John Wiley & Sons Ltd     

Multiclass discrimination of cervical precancers using Raman spectroscopy

Raman spectroscopy has the potential to differentiate among the various stages leading to high‐grade cervical cancer such as normal, squamous metaplasia, and low‐grade cancer. For Raman spectroscopy to successfully differentiate among the stages, an applicable statistical method must be developed. Algorithms like linear discriminant analysis (LDA) are incapable of differentiating among three or more types of tissues. We developed a novel statistical method combining the method of maximum representation and discrimination feature (MRDF) to extract diagnostic information with sparse multinomial logistic regression (SMLR) to classify spectra based on nonlinear features for multiclass analysis of Raman spectra. We found that high‐grade spectra classified correctly 95% of the time; low‐grade data classified correctly 74% of the time, improving sensitivity from 92 to 98% and specificity from 81 to 96% suggesting that MRDF with SMLR is a more appropriate technique for categorizing Raman spectra. SMLR also outputs a posterior probability to evaluate the algorithm's accuracy. This combined method holds promise to diagnose subtle changes leading to cervical cancer. Copyright © 2008 John Wiley & Sons, Ltd.     

Ⅱ型糖尿病人与大鼠红细胞拉曼光谱学比较

使用激光共聚焦拉曼光谱仪测量正常大鼠红细胞、正常人红细胞、糖尿病STZ造模大鼠红细胞、糖尿病四氧嘧啶造模大鼠红细胞和人Ⅱ型糖尿病红细胞的拉曼光谱,应用主成分分析(principal component analysis,PCA)结合支持向量机(support vector machines,SVM)分类器对数据进行判别分析,然后采用类间距离判断两种造模方法与人Ⅱ型糖尿病的接近程度。结果发现糖尿病红细胞与正常红细胞的拉曼光谱存在明显差异,糖尿病在酰胺 ⅥCO变形振动谱带处峰高显著,并在酰胺ⅤN—H变形振动谱带处谱线出现偏移,属于磷脂的脂酰基C—C骨架1 130 cm-1谱线增强,1 088 cm-1谱线强度减弱,说明糖尿病红细胞膜的通透性增强。PCA结合SVM可以很好地区分以上5类红细胞的拉曼光谱,分类器测试结果表明分类准确度达100%。通过分别计算两种造模方法与人Ⅱ型糖尿病的类间距离,发现STZ造模法更接近人Ⅱ型糖尿病。由此得出结论：拉曼光谱法可以用于糖尿病诊断,大鼠糖尿病STZ造模法更接近人类Ⅱ型糖尿病。     

Raman spectroscopy at different excitation wavelengths (1064, 785 and 532 nm) as a tool for diagnosis of colon cancer

Raman spectroscopy is structure sensitive non‐destructive method that allows observing the status of biological tissues with minimal impact. This method has a great potential in the diagnosis of various types of degenerative diseases including cancer damages. Near‐infrared Fourier transform (NIR‐FT)‐Raman (λ_ex ~1064 nm), NIR‐visible (Vis)‐Raman (λ_ex ~785 nm) and Vis‐Raman (λ_ex ~532 nm) spectra of normal and colorectal carcinoma colon tissue samples were recorded in macroscopic mode at 10–20 randomly chosen independent sites. In the cases of NIR‐Vis‐ and Vis‐Raman spectra, enhanced resonance effects were observed for tissue chromophores absorbing in the visible area. Evident spectral differences were noticed for Raman spectra of normal colon tissue samples in comparison with abnormal samples. The average Raman spectra of colon tissue samples were analysed by principal component analysis (PCA) to discriminate normal and abnormal tissues. PCA of combined dataset containing Raman intensities of chosen NIR‐FT, NIR‐Vis or Vis‐Raman bands led to discrimination of normal and abnormal colon tissue samples. Therefore, combination of these three Raman methods can be helpful for recognizing cancer lesions in colon for diagnostic purposes. Copyright © 2014 John Wiley & Sons, Ltd.     

Comparison of molecular images as defined by Raman spectra between normal mucosa and squamous cell carcinoma in the oral cavity

Raman spectroscopy has been effectively applied to clinically differentiate normal and cancerous mucosal tissues. Micro‐Raman spectroscopy provides a tool to better understand the molecular basis for the Raman clinical signal. The objective of the current study was to utilize micro‐Raman spectroscopy to define the molecular/spectral differences between normal and abnormal squamous cell carcinoma (SCC) in oral mucosa (in vitro). Understanding this may help in identifying unique spectra or may be useful for in vivo application of this technology. Micro‐Raman (confocal) spectroscopy was used to obtain molecular images of normal and SCC cells of human oral mucosa. Four fresh flashed‐frozen tumor and four matched normal tongue specimens were studied. The spectra covered a wavenumber range from 300 to 4000 cm⁻¹ with a spectral resolution of 8 cm⁻¹ and a spatial resolution of 1.0 µm. The cells were located within thin sections of tongue mucosa biopsies. The excitation wavelength of 515 nm was used. We were able to obtain Raman images with rich information about the spectroscopic and structural features within the cytoplasm, cell membrane, and cell nuclei. Significant spectral differences were observed between the Raman images of normal and malignant squamous cells. The heterogeneity of tumor cells within the abnormal tissue was also demonstrated. Spectral differences demonstrated between both tissue types have provided important information regarding the origins of specific signals within the cells of each tissue type. In our search for specific spectral biomarkers, we believe that a cell surface protein, greatly upregulated in SCC cells, was discovered at 1583 cm⁻¹. Copyright © 2010 John Wiley & Sons, Ltd.     

Raman spectroscopic identification of normal and malignant hepatocytes

Raman spectroscopy has strong potential for providing non-invasion diagnosis of cancers. In this paper, micro-Raman spectroscopy is used to diagnose one most common liver cancer, hepatocellular carcinoma （HCC）. The statistical analyzes, including t-test, principal component analysis （PCA）, and linear discriminant analysis （LDA）, are performed on the Raman spectra of malignant and normal hepatocytes. The t-test-LDA results show that the 786- and 1004-cm^-1 bands of the malignant and normal hepatocytes are significantly different, and PCA-LDA results show an overall accuracy of 100% for the Raman spectroscopic identification of normal and malignant hepatocytes in our experiment.     

Simultaneous Raman micro–spectroscopy of optically trapped and stacked cells

The combination of Raman spectroscopy and optical trapping holds great promise for single‐cell studies and is an emergent theme in microfluidic environments. Here, the evolution of the Raman signal intensity with an axial increment of the mass of the substance of interest inside a specific Raman excitation volume is investigated. Whilst Raman spectroscopy may be applied to tissue samples, solutions and single cells, there are no easily available methods to rapidly acquire signals from small cell populations. We show a simple but powerful method to record the Raman intensity signal simultaneously from a small number of trapped cells or colloidal particles using the technique of optical stacking. The Raman spectra of stacks of red blood cells and yeast cells show that this method can be applied to biological systems. We demonstrate how we may reveal biochemical fingerprints that would otherwise require long integration times for each single cell or averaging over many sequentially acquired cell spectra. There is potential to apply this method to directly attain Raman spectra from sorted sub‐populations of normal, abnormal and tumour cell lines. Copyright © 2007 John Wiley & Sons, Ltd.     

Evaluation of Raman spectroscopy for prediction of antitumor response to silibinin and its nanoparticulates in DMBA‐induced oral carcinogenesis

Raman spectroscopy is a vibrational spectroscopic technique that can be used to monitor the therapeutic efficacy of anticancer drugs during carcinogenesis in a non‐invasive and label‐free manner. The present study aims to investigate the biochemical changes exerted upon free silibinin (SIL) and its nanoparticulate (SILNPs) treatment against 7,12‐dimethylbenz[a]anthracene (DMBA)‐induced oral carcinogenesis in the fingerprint region of 1800–500 cm⁻¹ using HE‐785 Raman spectrometer. Raman spectra differed significantly between the control and tumor tissues, with tumor tissues characterized by increased intensities of vibrational bands such as nucleic acids, phenylalanine and tryptophan and a lower percentage of lipids when compared to the control tissues. Further, oral administration of free SIL and SILNPs significantly increased lipids and decreased the levels of tryptophan, phenylalanine and nucleic acid contents. Overall, the treatment of nanoparticulate SIL was found to be a more potent antitumor effect than free SIL in preventing the formation of tumor and also brought back the several Raman bands to a normal range in the buccal mucosa of hamsters during DMBA‐induced oral carcinogenesis. In addition, the detailed secondary structure of proteins in the control and experimental groups is also presented. Furthermore, the diagnostic algorithms based on principal component linear discriminant analysis (PC‐LDA) achieved an overall sensitivity of 94–100% and specificity of 76–100%. These results further demonstrate that Raman spectroscopy associated with PC‐LDA diagnostic algorithms could be a valuable tool for rapid and sensitive detection of specific biomolecular changes at the molecular level in response to anticancer drug. Copyright © 2015 John Wiley & Sons, Ltd.     

Multivariate analysis of combined Raman and fibre‐optic reflectance spectra for the identification of binder materials in simulated medieval paints

The non‐invasive identification of paint materials used in works of art is essential, both for preserving and restoring them, and also for understanding and verifying the history surrounding their creation. As such, the development of suitable non‐invasive techniques has received much interest in recent years. We have investigated the use of Fourier transform (FT)‐Raman spectroscopy and fibre‐optic reflectance spectroscopy (FORS), together with multivariate principal‐component analysis (PCA) techniques, in order to identify the pigment and binding materials used in made‐up samples representative of real artwork. We demonstrate that both types of spectroscopy provide complementary information which can be used to identify the pigments and binders in paint samples. We show that PCA with FT‐Raman spectra can be used to assist in the identification of oil‐based binders, and that the additional data provided by FORS spectra enables PCA on combined spectra to identify more complex proteinaceious and polysaccharide‐based binding media. The results presented here demonstrate that multivariate analyses of lead‐based paints, using data measured by FT‐Raman and FORS in conjunction, have much potential for identifying individual pigments and binders in paint samples. This provides a path towards computer‐assisted characterisation of paint materials on artwork. Copyright © 2013 John Wiley & Sons, Ltd.     

人胚胎干细胞与急性早幼粒细胞白血病细胞的拉曼光谱特征研究

急性早幼粒细胞白血病(APL)属于急性髓系白血病(AML),是FAB分型中的M3亚型。部分APL患者形成早幼粒细胞白血病/维甲酸受体融合基因,即PML-RARα融合基因。在内外界多种因素的共同作用下,早幼粒细胞白血病发病。胚胎干细胞(ESCs)具有多向分化的能力,在一定诱导条件下, ESCs可以向造血系统分化。早幼粒细胞位于ESCs分化下游,为粒系分化阶段的一种细胞。探索一种非标记的技术方法鉴别不同分化阶段造血细胞具有重要的科研和实践意义。拉曼光谱技术可用于多种类型疾病的鉴别诊断研究,近年来应用前景愈加广阔。实验研究人胚胎干细胞系(ES)、急性早幼粒细胞白血病细胞系(NB4)和急性早幼粒细胞白血病患者(M3)白血病细胞的拉曼光谱特征,建立拉曼光谱非标记鉴别不同分化阶段白血病的方法,为临床实验研究提供基础。分别收集胚胎干细胞系(ES)、急性早幼粒细胞白血病细胞系(NB4)和4例M3患者白血病细胞,使用Horiba Xplora拉曼光谱仪获取拉曼光谱,每组或每例患者采集25~30个白血病细胞光谱。结合应用主成分分析法(PCA)、判别函数分析(DFA)、系统聚类分析和偏最小二乘判别分析(PLS-DA),对三类细胞的光谱进行分析并建立模型,进而对三类细胞进行鉴别,应用交互验证法对模型进行验证。同时结合细胞超微结构分析三种细胞的拉曼光谱特征。M3, NB4和ES细胞的拉曼光谱差别显著,主要表现为M3和NB4细胞光谱中对应核酸、蛋白质及脂类物质的谱峰明显高于ES细胞,其生物学机制包含了APL与PI3K/Akt/mTOR通路的密切关系。PI3K/Akt/mTOR通路在急性早幼粒细胞白血病细胞中存在异常激活,影响白血病细胞的生物大分子代谢;鉴别建模的总体分类准确率达100%(181/181),交互验证的分类准确率达98.9%(179/181),表明鉴别模型预测能力良好。拉曼光谱分析显示M3细胞和NB4细胞增殖代谢明显高于ES细胞,根据PCA-DFA、聚类分析及PLS-DA建立的拉曼光谱鉴别模型能够准确区分3种不同分化阶段白血病相关细胞,其结果与电镜结果相符。     

Micro-Raman spectroscopy of single leukemic cells

The Raman spectra from leukemic cell line （HL60） and normal human peripheral blood mononuclear cells （PBMCs） are obtained by confocal micro-Raman spectroscopy using near-infrared laser （785 nm） excitation. The scanning range is from 500 to 2000 cm^-1. The two average Raman spectra of normal PBMCs and carcinoma cells have clear differences because their structure and amount of nucleic acid, protein, and other major molecules are changed. The spectra are also compared and analyzed by principal component analysis （PCA） to demonstrate the two distinct clusters of normal and transformed cells. The sensitivity of this technique for identifying transformed cells is 100%.     

广义模糊K调和均值聚类的近红外光谱生菜储藏时间鉴别

生菜的储藏时间是影响生菜新鲜程度的重要因素。为了快速、无损和有效地鉴别生菜的储藏时间,以欧式距离的p次方代替模糊K调和均值聚类(FKHM)中欧式距离的平方提出了一种广义模糊K调和均值聚类(GFKHM)算法并将该算法应用于鉴别生菜的储藏时间。以60个新鲜生菜样本为研究对象,采用Antaris Ⅱ近红外光谱分析仪每隔12 h检测生菜的近红外漫反射光谱,共检测三次,光谱扫描的波数范围为10 000~4 000 cm-1。首先用主成分分析(PCA)对1 557维的生菜近红外光谱进行降维处理以减少冗余信息,取前20个主成分,经过PCA处理后得到20维的数据。然后用线性判别分析(LDA)提取光谱数据的鉴别信息以提高聚类的准确率,取鉴别向量数为2,则LDA将20维的数据转换为2维数据。最后以模糊C-均值聚类(FCM)的类中心作为FKHM和GFKHM的初始聚类中心,分别运行FKHM和GFKHM计算模糊隶属度以实现生菜储藏时间的鉴别。结果表明,GFKHM的鉴别准确率能达到92.5%,FKHM的鉴别准确率为90.0%,GFKHM具有比FKHM更高的鉴别准确率。GFKHM的聚类中心比FKHM更逼近真实类中心。GFKHM的收敛速度明显快于FKHM。采用近红外光谱技术同时结合GFKHM,PCA和LDA为快速和无损地鉴别生菜储藏时间提供了一种新的方法。