Study On Differentiation Grades Of Hepatocellular Carcinoma Based On Multiphoton Microscopy

China is a country with high liver disease,with an annual incidence of 14.6%-46%.The incidence of primary liver cancer is as high as 25.7/100000,and it is increasing year by year.About 280,000 people die of liver cancer each year,accounting for 53% of the world's deaths.To date,the diagnosis and treatment of liver diseases,especially the diagnosis and treatment of liver cancer,remains challenging.Hepatocellular carcinoma(HCC)accounts for more than 90% of the total incidence of liver cancer.HCC patients can have distinct therapeutic responses and overall effects depending on the grade of differentiation of the tumor.Therefore,the clear and accurate classification of HCC differentiation has great clinical value.At present,HCC differentiation is difficult to classify.It only depends on the expertise of the pathologist and the time required for diagnosis.Despite combining different indicators,the researchers developed a variety of grading and staging systems for selecting treatment options and predicting prognosis.However,the traditional grading system was originally developed for liver disease.More attention is paid to the qualitative rather than quantitative characteristics of cancer development,such as Child-Pugh scoring system,MELD evaluation system(end-stage liver disease model),CLIP scoring method,Oknda staging system,and TNM staging system.Therefore,it is of great clinical significance to establish a prediction model based on the characteristics of collagen fibers.Multiphoton microscopy(MPM)is an emerging nonlinear optical imaging technology by detecting optical signals such as two-photon excitation fluorescence and second harmonic generation produced by the interaction of femtosecond laser with internal components of biological tissue.It has high spatial resolution,low cell damage,large imaging depth and can be applied to tissue three-dimensional imaging.The second harmonic generation(SHG)signal and two-photon excitation fluorescence(TPEF)signal collected by MPM provide additional quantitative information and specificity not found in conventional pathology.MPM has been widely used in basic research and clinical research in the liver,and has become an important topic in the field of biomedical photonics.However,the application of MPM to the differentiation of HCC is still unknown.Based on this,this paper uses MPM to conduct in-depth research on HCC with different differentiation grades,including the development of LSIE-MPM algorithm for the weak signal of HCC samples,and classification of multiphoton images of HCC withdifferent differentiation,and automatically classification of different differentiation grades of HCC by the convolutional neural network(CNN),and the predictive model for the well,moderately and poorly differentiated grades and prognosis of HCC.The main contents are summarized as follows:Firstly,in order to overcome the shortcomings of weak endogenous signals in liver tissue,this paper combines the chi-square transformation function and the luminance transformation function model to develop a low-signal enhancement algorithm(LSIE-MPM).This method was first applied to MPM imaging and validated on different types of MPM images,such as plant cells,fresh ovarian tissue and real-time video.It has finally applied to signal enhancement of multiphoton images of HCC with different differentiation.The results show that LSIE-MPM significantly enhances the signals of multiphoton images,and the effective information is clearly visible.At the same time,laser damage to the sample is avoided.The main results related to this section have been published in Journal of Physics D: Applied Physics,2019,52: 285401(SCI-II,IF =2.829).Secondly,in this paper,MPM was applied to paraffin-embedded samples of HCC.The label-free classification of well,moderately and poorly differentiation grades was carried out.After feature extraction and feature selection,Mann-Whitney test and receiver operating characteristic curve analysis were applied to the SHG signals from collagen inside tumors.The results showed a good correlation between software analysis and the diagnosis of experienced pathologists.Combining image features and clinical information,an adaptive quantification algorithm was generated to automatically determine HCC differentiation levels.The results suggest that MPM may be a promising automated diagnostic method for clinical use without the need for time-consuming of tissue processing and staining.The main results related to this section have been published in BIOMEDICAL OPTICS EXPRESS,2018,9(8): 3783-3793(SCI-II,IF =3.91).Further,we have studied on the fusion of MPM and CNN algorithms to distinguish the differentiation level of HCC to produce innovative computer-aided diagnostic methods.Our model trained the composite images containing TPEF signal and SHG signal based on the VGG-16 framework.The resulting classification accuracy of HCC differentiation grades is higher than 90%.The results show that MPM and CNNs can be combined to implement a label-free automated method for various tissue,disease and other related classification problems.The main results related to this section have been published in JOURNAL OF BIOPHOTONICS,2019,12(7): e201970024(SCI-II,IF=3.763).Finally,in order to prove that the differentiation grades of HCC is closely related to the prognosis,this paper combined the changes of collagen characteristics inside the tumor with two important prognostic indicators,extrahepatic metastasis and recurrence,and established a multivariate logistic regression model: Collagen Score(CS).The CS model combined the collagen characteristics under MPM with clinical information such as the differentiation grades.The prediction accuracy of extrahepatic metastasis and recurrence was 95.3% and 71.6%,respectively.The results indicate that collagen fibers serve as a stable and marker-free prognostic marker.The main result related to this section "Collagen Score: A Novel Method for Predicting the Prognosis of Liver Cancer" is being submitted.In summary,through the research of this paper,the research gap of MPM technology in the differentiation grades of HCC has been filled.A new computer-aided diagnosis method based on MPM has been developed.