乳腺组织的拉曼光谱线性模型研究

本研究甄选10个代表乳腺组织拉曼活性成分的基谱,在国内首次构建一种乳腺组织拉曼谱的线性回归模型。用2 000多个正常和非正常乳腺组织拉曼谱对该模型进行统计检验,模型显著性F检验的置信度全部为1,多元决定系数的平均值为0.95,表明线性模型假设成立、拟合效果良好。10个基谱代表脂肪、细胞质、细胞间质、DNA、血液、β-胡萝卜素、胆固醇等的拉曼谱,基谱的归一化拟合系数间接反映出这些成分的相对含量。用该模型拟合正常和肿瘤乳腺组织的拉曼宏观谱,病变前后细胞质和DNA的拟合系数增大、脂肪拟合系数减小,这反映出它们相对含量的增减,与已知的病理学结果一致。该研究有助于理解乳腺肿瘤组织的生化变化,并可能为分析该变化提供了有效手段。

Study on Raman Linear Model of Human Breast Tissue

A linear regression model for Raman spectra of human breast tissue was developed with 10 basis spectra which were derived from the spectra of fat, cell cytoplasm, collagen, beta-carotene, DNA and cholesterol etc. This model was tested statistically with more than 2 000 spectra collected from both normal and tumor breast tissue samples. The confidence of F-test for the model significance was 1 to all the spectra and the average of multiple determination coefficients measuring the fit goodness was 0.95. The normalized fit coefficients reflected, to a certain extent, the relative quantities of the Raman-active components in tissue represented by the basis spectra. In comparison with normal breast tissue, the coefficients of the cell cytoplasm and DNA basis spectra increased apparently for the tumor one, and the coefficient of fat basis spectrum changed reversely, which was easily understood with the known pathological changes in the tissue. Mapping spectra were also fitted with this model. This work is helpful to understanding the biochemical/morphological changes in the breast tumor tissues and to developing the Raman diagnostic method for breast tumor.