Establishment Of Diagnostic Model Of Gastric Cancer By Raman Spectroscopy And Analysis Of Its Raman Spectrum Characteristics

Background and ObjectiveGastric cancer is a common disease which can have a good prognosis and even be cured by early treatment.However,early diagnostic of gastric cancer has become a major problem.Traditional diagnostic methods rely on biopsy which shows time-consuming,bleeding,re-fetch and other shortcomings.New diagnostic methods,such as NBI(Narrow banding image)and magnifying endoscopy,are depended on the doctors' experience,and needed to learn for a long time,which is affected its clinical application to some extent.Raman spectroscopy reflects the molecular composition and molecular structure of various substance.It has the advantage of non-destructive,rapid,without interference of water.The development of this technique in the biomedical field are rapid in recent years.Human body is consist of biological macromolecules.Cancer is starting from the molecular alterations,and gradually develop into tissue alterations.Molecular alterations of gastric cancer can be found by Raman spectroscopy at early stage.Raman spectroscopy can be divided into the fingerprint region(FP 800-1800cm-1)and high wavenumber region(HW 2800-3700cm-1).The Raman peaks of FP region are intensive.The biochemical information of molecular in FP region are rich and have been reported in a number of related research works.The HW region is not easy be interfered by fluorescence and is seldom reported.The aim of this study is to analyze the Raman spectrum features of gastric cancer in FP region and HW region,and establish the diagnostic model to evaluate the diagnostic potential of Raman spectrum.MethodUsing confocal Raman microscope to collect the spectral of normal gastric mucosa and gastric cancer tissues.The parametric Student's t-test were performed to compare differences between the two groups.A p-value less than 0.05 was considered to indicate statistical significance.Probabilistic partial least-squares(PLS)-discriminant analysis(DA),was applied to evaluate the difference between normal gastric mucosa and gastric cancer.The leave-one-sample-out cross-validation method was used to confirm the performance of the PLS-DA diagnostic algorithm.The effect of the diagnostic models(FP ? HW ?FP/HW)was evaluated by the area under the ROC curve.Results1.Raman peaks of gastric cancer were lower than the normal tissues at 829 cm-1,853 cm-1?879 cm-1,898 cm-1,1003 cm-1,1033 cm-1,1047 cm-1,1065 cm-1,1270 cm-1,1448 cm-1,1658 cm-1,2932cm-1,and at 1157 cm-1,1173cm-1,1304cm-1,1319cm-1,1338cm-1,1374cm-1,1476 cm-1,1526 cm-1,1553 cm-1,1583 cm-1 were higher than the normal.2.Sensitivity of the diagnosis model of the FP region,HW region and FP/HW region were 94.59%?81.08%?94.59%,specificity were 86.84%?71.05%?86.84 %,area under the ROC curve were 0.9395?0.8001?0.9264ConclusionThe FP region and the HW region of normal gastric mucosa and gastric cancer contained a numerous specific peak information.The peaks of two regions complement each other and support each other.The effects diagnostic model of FP and the FP region/HW region were superior to HW region.Raman spectroscopy has excellent skills of qualitative analytical,and have a good prospect in the clinical diagnosis of gastric cancer.