Study On Infrared And Raman Spectra Of Serum From Patients With Liver Cancer And Lung Cancer

Liver cancer and lung cancer are two fatal malignant tumors threatening human health.Traditional tumor marker methods and existing imaging methods have some limitations for their early diagnosis.In this dissertation,Fourier transform infrared spectroscopy（FTIR）and Raman spectroscopy were used to study the serum from patients with liver cancer,patients with lung cancer and healthy people,in order to explore an economical,wide applicable and high-accuracy diagnostic method for liver cancer and lung cancer,laying a foundation for the clinical application of infrared spectroscopy and Raman spectroscopy in the diagnosis of liver cancer and lung cancer in the future.The main works are follows:1.The Fourier transform infrared spectra of the serum from patients with liver cancer and healthy people were similar on the whole.The curve fitting analysis was performed on the infrared spectra of the serum in the protein range of 1700-1600 cm^-1,and it was found that there were differences in secondary structure of protein.Principal component analysis（PCA）and partial least squares discriminant analysis（PLS-DA）were performed on the second derivative infrared spectra（SD-IR）in the lipid range of2900-2800 cm^-1.The results of principal component analysis showed that the serum samples from patients with liver cancer were well classified from those from healthy people,and the first two principal components explained 95%of the total data variance;the results of partial least squares discriminant analysis showed 92.85%sensitivity and95.23%specificity,and the total accuracy was 94.29%.2.The attenuated total reflection infrared spectra（ATR-FTIR）of serum samples showed that compared with the serum from healthy people,the average infrared spectra of serum from patients with lung cancer was increased significantly in the nucleic acid range of 1250-1000 cm^-1.The results of band area analysis showed that the concentrations of protein,lipid and nucleic acid molecules in the serum from patients with lung cancer were increased compared with those from healthy people.Partial least squares discriminant analysis of the first derivative spectral data in the nucleic acid range of 1250-1000 cm^-1 showed 80%sensitivity and 91.89%specificity,with a total accuracy of 87.10%.3.Raman spectra were collected from serum samples from patients with liver cancer and healthy people,and the changes of carotene,protein and lipid in human serum during the occurrence of liver cancer were reflected by comparing the intensity differences of Raman spectral characteristic peaks.The combination of principal component analysis and partial least squares discriminant analysis could exactly classify limited serum samples from patients with liver cancer and those from healthy people.4.Wavelet denoising and data fusion technique combined with infrared spectroscopy and Raman spectroscopy were used to study the serum samples from patients with lung cancer and healthy people.The results showed that Raman spectroscopy can provide more biological information than infrared spectroscopy for serum samples.Wavelet threshold denoising（WTD）can filter the invalid information from the original spectral data.After selecting the best wavelet parameters,the fused data after wavelet denoising showed 96.08%sensitivity and 90%specificity in the validation set of partial least squares discriminant analysis model,and the total accuracy rate could reach 93.41%.The work of this dissertation shows that by studying the infrared spectra and Raman spectra of serum,combined with second derivative spectroscopy,curve fitting analysis,chemometrics analysis,wavelet denoising and data fusion analysis,patients with liver cancer and patients with lung cancer could be diagnosed quickly and accurately.