拉曼光谱在鉴别垂体腺瘤分泌类型中的应用  

Application of laser confocal micro-Raman spectroscopy in identification of the secretory types of pituitary adenomas

作　　者：刘杰[1,2,3,4] 汪攀 汤超 张华 吴南[1,2,3,4] Liu Jie;Wang Pan;Tang Chao;Zhang Hua;Wu Nan(Chongqing Medical University,Chongqing 400016,China;Chongqing Institute of Green and Intelligent Technology,Chinese Academy of Sciences,Chongqing 400714,China;Chongqing School,University of Academy of Sciences,Chongqing 400714,China;Department of Neurosurgery,Chongqing General Hospital,Chongqing 401147,China)

机构地区：[1]重庆医科大学,重庆400016 [2]中国科学院重庆绿色智能技术研究院,重庆400714 [3]中国科学院大学重庆学院,重庆400714 [4]重庆市人民医院神经外科,重庆401147

出　　处：《中华神经外科杂志》2023年第4期356-361,共6页Chinese Journal of Neurosurgery

基　　金：重庆市科学技术局基金(CSTC2021jscx-gksb-N0024);重庆市人民医院院级课题(2020.6.4)。

摘　　要：目的探讨应用激光共聚焦显微拉曼光谱结合不同的分析方法构建的模型在鉴别垂体腺瘤(PA)分泌类型中的作用。方法收集2020年12月至2022年4月重庆市人民医院神经外科手术切除的20例PA患者的肿瘤标本。术后经病理学证实促性腺激素(Gn)细胞腺瘤、生长激素(GH)细胞腺瘤、零细胞腺瘤各5例,催乳激素(PRL)细胞腺瘤4例,促肾上腺皮质激素(ACTH)细胞腺瘤1例。应用激光共聚焦显微拉曼光谱仪检测不同分泌类型PA的拉曼光谱。将4/5的光谱数据作为训练集,1/5的光谱数据作为测试集。采用主成分分析(PCA)联合线性判别分析(LDA,PCA-LDA)或二次判别分析(QDA,PCA-QDA)法分析训练集光谱数据并构建定性判别模型。利用测试集数据分析模型判定PA分泌类型的准确率。结果共获得2000条拉曼光谱,其中Gn细胞腺瘤500条,PRL细胞腺瘤400条,GH细胞腺瘤500条,零细胞腺瘤500条,ACTH细胞腺瘤100条。PRL细胞腺瘤的光谱在1064、1128、1342 cm^(-1)出现明显的峰位移,GH细胞腺瘤的光谱在1004 cm^(-1)出现明显的峰位移,Gn细胞腺瘤的光谱在854、1660 cm^(-1)出现明显的峰位移,ACTH细胞腺瘤的光谱在854、1128、1451 cm^(-1)出现明显的峰位移,零细胞腺瘤的光谱在1004、1660 cm^(-1)出现明显的峰位移。当结合光谱数据的前8个主成分进行分析时,PCA-LDA构建的模型鉴别PA分泌类型的准确率为76.61%,PCA-QDA模型的准确率为96.64%。结论不同分泌类型PA的拉曼光谱存在差异;利用激光共聚焦显微拉曼光谱结合PCA-QDA构建定性判别模型可以有效鉴别不同分泌类型的PA。Objective To explore the role of laser confocal micro-Raman spectroscopy combined with different analytical methods in identification of the secretory types of pituitary adenomas(PAs).Methods The tumor samples of 20 PA patients surgically resected in the Department of Neurosurgery of Chongqing General Hospital from December 2020 to April 2022 were collected.Postsurgical pathological examinations revealed that there were 5 cases of gonadotropin cell adenoma,5 cases of growth hormone(GH)cell adenoma,5 cases of zero-cell adenoma,4 cases of prolactin(PRL)cell adenoma,and 1 case of adrenocorticotropin hormone(ACTH)cell adenoma.Laser confocal micro-Raman spectroscopy was applied to detect the Raman spectra of different secretory types of PAs.We selected 4/5 of the obtained spectral data as the training set and 1/5 of the spectral data as the test set.The principal component analysis(PCA)combined with linear discriminant analysis(LDA,PCA-LDA)or quadratic discriminant analysis(QDA,PCA-QDA)was used to analyze the training set spectral data and construct qualitative discriminant models.The accuracy of the model based on the test set was analyzed to determine the types of PA secretion.Results A total of 2000 Raman spectra were obtained,including 500 for gonadotropin cell adenomas,500 for GH cell adenomas,500 for zero-cell adenomas,400 for PRL cell adenomas,and 100 for ACTH cell adenomas.The spectra of PRL cell adenoma showed significant peak shifts at 1064,1128,and 1342 cm^(-1),the spectra of GH cell adenoma showed significant peak shifts at 1004 cm^(-1),the spectra of gonadotropin cell adenoma showed significant peak shifts at 854 and 1660 cm^(-1),the spectra of ACTH cell adenoma showed significant peak shifts at 854,1128,and 1451 cm^(-1),and the spectra of zero-cell adenoma showed significant peak shifts at 1004 and 1660 cm^(-1).The PCA-LDA and PCA-QDA qualitative discrimination models were constructed after analysis of the training set of Raman spectral data.When combining the first 8 principal components of the spectral

关 键 词：垂体肿瘤 光谱分析 拉曼 主成分分析 线性判别分析 二次判别分析 

分 类 号：R736.4[医药卫生—肿瘤]