Minimally invasive cancer detection using bio-fluids is been actively pursued due to practical limitations of, though, better suited noninvasive and online in vivo methods. Saliva is one such clinically informative bio-fluid that offers advantage of easy and multiple sample collection. Despite its potential in cancer diagnostics, saliva analysis is challenging due to its heterogeneous composition. Recently, there has been an upsurge in saliva exploration by optical techniques. Although forms of saliva such as precipitate and supernatant have been monitored, but due to obvious loss of analytes in processing, sampling method needs to be standardized. In that context, present work details the comparison of four different saliva sampling methodologies i.e., air-dried (D), lyophilized (L), pellet (P) and supernatant (S) by Raman spectroscopy (RS) collected from 10 healthy samples. Composition driven spectral features of all forms were compared and classified using principal component analysis (PCA) and linear discriminant analysis (PCLDA). Analysis was carried out on all four groups in the first step. In the second step groups of P and S; and D and L were analyzed. Findings suggest that P and S exhibit discrete spectroscopic features and demonstrated high classification efficiency which is indicative of their distinctive biochemical composition. On the other hand, D and L forms showed overlapping spectral features and low classification suggesting these forms retain majority spectroscopic features of whole saliva and less prone to sampling losses. Thus, this study indicates D and L forms may be more appropriate for saliva sampling by RS providing the comprehensive information, required for cancer diagnosis. Further the method was also tested for classification of oral cancer and healthy subjects (n= 27) which yields 90 % stratification. Thus the findings of the study indicate utility of minimally invasive salivary Raman based diagnostic techniques in oral cancers.

中文翻译：

---

EXPRESS：唾液拉曼光谱：健康和口腔癌受试者的采样方案和分层的标准化

然而，由于更适合无创和在线体内方法的实际限制，使用生物流体的微创癌症检测得到了积极的追求。唾液就是这样一种临床信息丰富的生物流体，它提供了简单和多样本采集的优势。尽管它在癌症诊断方面具有潜力，但唾液分析由于其异质成分而具有挑战性。最近，通过光学技术进行唾液探索的热潮。虽然对沉淀物、上清液等唾液形态进行了监测，但由于处理过程中分析物损失明显，采样方法需要标准化。在这种情况下，目前的工作详细介绍了四种不同唾液采样方法的比较，即风干 (D)、冻干 (L)、通过拉曼光谱 (RS) 从 10 个健康样品中收集的颗粒 (P) 和上清液 (S)。使用主成分分析 (PCA) 和线性判别分析 (PCLDA) 比较和分类所有形式的成分驱动的光谱特征。在第一步中对所有四组进行了分析。在 P 和 S 的第二步组中；和 D 和 L 进行了分析。研究结果表明 P 和 S 表现出离散的光谱特征，并表现出高分类效率，这表明它们独特的生化成分。另一方面，D 和 L 形式显示出重叠的光谱特征和低分类，表明这些形式保留了整个唾液的大部分光谱特征​​，并且不太容易发生采样损失。因此，这项研究表明 D 和 L 型可能更适合通过 RS 进行唾液采样，提供癌症诊断所需的综合信息。此外，该方法还针对口腔癌和健康受试者（n=27）的分类进行了测试，产生了 90% 的分层。因此，该研究的结果表明基于微创唾液拉曼的诊断技术在口腔癌中的实用性。