Optical Bioimaging Laboratory, Department of Biomedical Engineering, Faculty of Engineering , National University of Singapore , 9 Engineering Drive 1 , Singapore 117576.
Department of Otolaryngology, Head and Neck Surgery , National University of Singapore and National University Health System , Singapore 119074.
Anal Chem. 2019 Jul 2;91(13):8101-8108. doi: 10.1021/acs.analchem.9b00173. Epub 2019 Jun 10.
Raman spectroscopy is an optical vibrational spectroscopic technique capable of probing specific biochemical structures and conformation of tissue and cells in biomedical systems. This work aims to assess the clinical utility of a fiber-optic Raman spectroscopy with nature-inspired genetic algorithms for enhancing in vivo detection and diagnosis of nasopharyngeal carcinoma (NPC) patients. The Raman diagnostic platform is developed based on simultaneous fingerprint (FP) and high-wavenumber (HW) fiber-optic Raman endoscopy associated with genetic algorithms-partial least-squares-linear discriminant analysis (GA-PLS-LDA). A total of 2126 in vivo FP/HW Raman spectra (598 NPC, 1528 normal) acquired from 113 tissue sites of 14 NPC patients and 48 healthy subjects during nasopharyngeal endoscopic examinations. Distinct Raman peaks have been identified (853 cm - proteins, 1209 cm - phenylalanine, 1265 cm - proteins, 1335 cm - proteins and nucleic acids, 1554 cm - tryptophan, porphyrin, 2885 cm - lipids, 2940 cm - proteins, 3009 cm - lipids, and 3250 cm - water) that are related to the significant biochemical changes ( p < 1 × 10) in NPC compared to normal tissue. Raman diagnostic performance is evaluated through the leave-one-object (tissue site)-out cross-validation (LOOCV) method. A statistically significant GA-PLS-LDA model ( p < 1 × 10) on FP/HW Raman yields a CV diagnostic accuracy of 98.23% (111/113), sensitivity of 93.33% (28/30), and specificity of 100% (83/83) for NPC classification. This work demonstrates that the fiber-optic FP/HW Raman diagnostic platform developed has great promise for improving real-time in vivo detection and diagnosis of NPC at the molecular level during clinical nasopharyngeal endoscopy.
拉曼光谱是一种光学振动光谱技术，能够探测生物医学系统中组织和细胞的特定生化结构和构象。本工作旨在评估一种基于光纤拉曼光谱和受自然启发的遗传算法的诊断平台，以提高对鼻咽癌（NPC）患者的体内检测和诊断的临床实用性。拉曼诊断平台是基于同时获得的指纹（FP）和高波数（HW）光纤拉曼内窥镜与遗传算法-偏最小二乘线性判别分析（GA-PLS-LDA）相关联而开发的。共采集了 14 名 NPC 患者和 48 名健康受试者在鼻咽内镜检查期间的 113 个组织部位的 2126 个体内 FP/HW 拉曼光谱（598 个 NPC，1528 个正常）。已经鉴定出明显的拉曼峰（853cm - 蛋白质，1209cm - 苯丙氨酸，1265cm - 蛋白质，1335cm - 蛋白质和核酸，1554cm - 色氨酸，卟啉，2885cm - 脂质，2940cm - 蛋白质，3009cm - 脂质，和 3250cm - 水），这些峰与 NPC 组织中与正常组织相比具有显著的生化变化（p < 1 × 10）有关。通过留一对象（组织部位）交叉验证（LOOCV）方法评估拉曼诊断性能。FP/HW 拉曼的统计学上显著的 GA-PLS-LDA 模型（p < 1 × 10）产生了 98.23%（111/113）的 CV 诊断准确率，93.33%（28/30）的灵敏度和 100%（83/83）的特异性用于 NPC 分类。这项工作表明，所开发的光纤 FP/HW 拉曼诊断平台具有很大的潜力，可以提高临床鼻咽内镜检查时 NPC 的实时体内检测和诊断的分子水平。
