Study On The Differential Diagnosis Of Non-hodgkin’s Lymphoma Using SERS Technique Based On Serum Samples

Non-Hodgkin’s lymphoma(NHL)is the most common cancer in the lymphatic system,in which rapid and accurate diagnosis is an important clinical challenge.Surface-enhanced Raman scattering(SERS),due to its advantages of non-invasiveness,high sensitivity,and no labeling,when combined with machine learning algorithms,is expected to become a powerful technical means for differential diagnosis of cancer.At present,there is no report on the diagnosis of non-Hodgkin’s lymphoma based on serum SERS technology.Based on this,this dissertation carried out the diagnosis,classification and staging of NHL based on serum SERS technology.Firstly,this dissertation introduces the research progress of SERS technology in the diagnosis and analysis of biomedical samples such as cells,tissues,and serum,analyzes the research status of the technology in cancer diagnosis,and proposes the research content of this dissertation.Secondly,the physical mechanism of SERS is introduced,and machine learning methods such as principal component analysis(PCA),k nearest neighbor(k NN),linear discriminant analysis(LDA),support vector machine(SVM),and feedforward neural network(FNN),as well as the evaluation methods of classification models such as cross-validation and confusion matrix are given.Thirdly,silver nano-particles for SERS enhancement are prepared and characterized,the detection substrate and the ratio of nanoparticles to serum is optimized,and the coffee ring effect of serum is analyzed.The SERS spectra of serum samples from patients with lymphoma and healthy controls are obtained using a semiconductor laser with a wavelength of 785 nm,and a database of serum SERS spectra is established.Comparative analysis of SERS spectra of serum samples of non-Hodgkin’s lymphoma patients and healthy people,diffuse large B cell lymphoma(DLBCL)subtype and follicular non-Hodgkin’s lymphoma(FNHL)subtype,different stages of DLBCL and DLBCL in germinal centers and outside germinal centers is performed.It is found that the intensity and width of Raman characteristic peaks of nucleic acids,sugars,phenylalanine,polypeptide proteins,and lipids in the above groups of spectra are significantly different.Finally,using the serum SERS spectral data,the differential diagnosis,classification and staging models of NHL are establishedbased on k NN,LDA,SVM and FNN.The models were trained and evaluated using 10-fold cross-validation.The results show that in the diagnosis and classification of NHL,FNN has the best performances,and the diagnostic accuracy is 93.1% and 85.9%,respectively.In the NHL staging application,the diagnostic accuracy of k NN is the highest,being 80.3%.The research results of this paper have important clinical value in the rapid and accurate diagnosis of non-Hodgkin’s lymphoma.