enhanced Raman spectroscopy of blood serum based on gold nan

1371 and 1445 cm−1. These peaks were assigned to the nucleic acids and proteins which could be caused by the active metabolism of nucleic acids in the patients with cancers [， ， there were the increasing intensities of peaks at 1139， ]. Rekha et al. [] reported that the plasma Raman spectra of the oral premalignant and malignant conditions were different from the normal condition， so there were some differences between their results and ours. Feng et al. [] reported that the SERS band at 725 cm−1 were greater in nasopharyngeal cancer plasma than the ones in the normal plasma， and could be utilized to diagnose the oral premalignant and malignant conditions with the sensitivity and specificity of above 80.0%. Serum Raman spectroscopy had the potential not only in the diagnosis of oral cancers，。

and the band was assigned to the nucleic acids. The increasing proteins in the serum of OSCC and MEC could result from the redistribution or translocation of free amino acids in the blood of the patients with cancers []. Compared with the MEC group， PCA was not suitable for classification problems because it could not use any class information in computing the features extracted from the original data []. So Belhumeur et al. [] combined PCA with LDA to make a popular method for dimension reduction and classification of data sets. LDA was a well-known method which found a linear transformation such that feature clusters were most seperable after the transformation []. The PCA combined with LDA were employed to classify and diagnose the different disease successfully based on the Raman spectral features [， and the only routine Raman spectra were obtained， ]. These findings were similar to the results reported in the previous studies. Rekha et al. [] and Sahu et al. [] all found that the increasing intensities of peaks at 1339 and 1445 cm−1 were related to the increase of the nucleic acids and proteins in the plasma and serum of the oral cancer groups. But in their reported studies， the spectral differences of the peaks assigned to various biomolecular fingerprints were captured in the SERS. Compared with the normal control group。

726， 1263， ]. The serum or plasma have been selected as the samples for the cancer detection and diagnosis by Raman spectroscopy successfully because the serum or plasma is easily collected and mostly contains some vital biomarkers generated by cancers [。

，745， –]. Among these algorithms， which showed a giant potential to diagnose OSCC based on the SERS. ， cell apoptosis and tumor necrosis， 548， ， the ranges of Raman shifts were limited from the 700 cm−1 or 800 cm−1 to 180 cm−1， ， numerous algorithms were employed to analyze Raman spectral data in the reported studies [。

but also in the prediction of treatment responses []. Accordingly， The American Academy of Oral Medicine recognized that the patients with oral cancer detected in the early stages required less aggressive treatment and experienced fewer complications than the ones with advanced stage cancers， MEC and normal groups with a total accuracy of 68.9%， ]. Rekha et al. [] used PCA-LDA to yield a diagnostic sensitivity of 91.2% and a specificity of 96.7% in the classification of normal from oral malignant group. In this study， the serum was the sample of the choice in this study. The SERS has been employed to detect and diagnose the cancers with a high accuracy in different sites. High-quality SERS of serum was used to identify the colorectal cancer with the diagnostic accuracy of 100% []. The SERS in this study successfully made the serum constituents adsorb on the nanoparticles and captured the biomolecular spectra fingerprints of the biomarkers. In this study， the PCA-LDA could also successfully classify and diagnose the spectra of OSCC and normal group with a sensitivity of 80.7% and a specificity of 84.1%. The results of the cross validation demonstrated the utilization of the PCA-LDA algorithm in the analysis of Raman spectral data was reliable. Then PCA-LDA had an excellent performance in the analysis and classification of the spectra of the OSCC。

and the patients with early stage oral cancer would have improved survival []. And the evidence existed in a systematic review that the development and use of biomarkers was becoming increasingly common in the early detection of oral cancer []. The metabolites in the biofluids of the patients with cancers are different from the normal subjects， which can be used for the cancer detection as the biomarkers [， the spectra of OSCC and MEC group exhibited the increasing intensities of the peaks at the 446 ~ 476。

due to the amino acid metabolism， PCA was a statistical technique for reducing the dimensions and simplifying complex data sets with the minimum reconstruction error []. However，1491 cm−1 and the decreasing intensity of peak at 1602 cm−1 in the SERS of OSCC group. These differences demonstrated the higher D-mannos and lipids level in the OSCC group. The previous study reported that there was an alteration in plasma lipid constituents in cancer patients because the cancer cells would utilize lipids for new membrane biogenesis []. So all above spectral differences captured in the SERS could be employed as the diagnostic indicators and references to detect the OSCC serum samples. In order to develop and establish the classification and diagnostic models。