The Reserch Of Fourier Transform Infrared Spectroscopy To Aiding Pathological Diagnosis Of Colorectal Cancer

Colorectal cancer(CRC)is one of the most common malignant tumor,early and accurate diagnosis can reduce the mortality of patients.Among the methods,pathological diagnosis is regarded as a method of sentence and authority.The golden standards,however,is not absolute authoritative and omnipotent with inherent limitations of subjective and objective.Fourier transform infrared(FT-IR)spectroscopy with the characteristics of easy to perform and suitable for the analysis of complex samples,can reveal the differences of cancerous and normal tissue from molecular level.Combine with chemometric techniques,it can achieve the accurate and rapid identification of colorectal cancer tissue category to aiding pathologic diagnosis.This research used mid-infrared(MIR)spectroscopy and near infrared(NIR)spectroscopy combined with discriminant analysis(DA)and counterpropagation artificial neural network(CP-ANN).Based on theses,the colorectal tissue slices for pathological diagnosis were analyzed to established the accurate and rapid method and provide technical support for pathological diagnosis.OBJIECTIVE:To establish the accurate and rapid IR method for the identification of colorectal tissue.That could be a promising candidate method to aiding pathological diagnosis of colorectal cancer.METHODS:1.The spectra measurement of colorectal tissue pathological slices.We use the Fourier transform infrared analyzer to measure the mid-infrared transmittance spectra(MIRTS),with the resolution of 4 cm-1,scans of 32 and spectral range of 4000 – 2200 cm-1.Fourier transform near infrared analyzer with diffuse reflectance accessory were used to measure the near infrared diffuse reflectance spectra(NIRDRS),with the resolution of 8 cm-1,scans of 64 and spectral range of 10000 – 4000 cm-1.2.Based on the MIRTS feature regions,DA and CP-ANN were investigated to establish the identify models of colorectal pathological slices.3.Based on the NIRDRS feature regions,DA and CP-ANN were investigated to establish the identify models of colorectal pathological slices.RESULTS:1.213 MIRTS and 213 NIRDRS were obtained from 71 colorectal tissues slices.2.By application the MIRTS of tissue slices,the established DA and CP-ANN models can distinguish the colorectal tissue slices accurately and quickly.The correction rate of calibration and validation were 99.31% and 100.0% in DA model respectively.In CP-ANN model,the correction rate of calibration set,prediction set and cross-validation were all 100.0%.3.By application the NIRDRS of tissue sections,the established DA and CP-ANN models can distinguish the colorectal tissue slices accurately and quickly.The correction rate of calibration and validation were 99.31% and 100.0% in DA model respectively.In CP-ANN model,the correction rate of calibration set,prediction set and cross-validation were all 100.0%.CONCLUTION:Based on the colorectal tissue slices for pathological diagnosis,the MIRTS and NIRDRS combined with linear DA and non-linear CP-ANN models can discriminant the slices accurately and quickly.The strategy could be used to aiding the pathological diagnosis of colorectal cancer.