Font Size: a A A

Raman Spectral Analysis Of Cancer Tissue Based On Support Vector Machines

Posted on:2016-08-24Degree:MasterType:Thesis
Country:ChinaCandidate:L B GuoFull Text:PDF
GTID:2308330473959876Subject:Optical Engineering
Abstract/Summary:PDF Full Text Request
With the fast development of Raman spectroscopy, those technologies have been used in the biomedical research. In this thesis, introduce the laser Raman spectroscopy in the cancer tissue detection research significance. In addition, we propose the support vector machine on classification of different cancer tissues according to the features of Raman spectroscopy. We also preliminary discuss the kernel functions and their parameters. The main research shows as follows:1. In order to solve the classification of Raman spectra and SERS data, we propose the SVM algorithm. Select the Raman spectra and SERS data of gastric cancer, hepatocellular carcinoma and nasopharyngeal cancer. According to the comparison of three kinds of kernel functions for SVM classification, we find that the RBF kernel SVM algorithms achieve optimal performance.2. For the same sample data, we contrasted the classifications between the RBF kernel SVM and LDA, the result showed that for the SERS data, SVM is superior to LDA, but for the Raman spectra data, SVM goes near to the performance of LDA.3. On account of general performance of SVM, we propose that optimize the parameters of RBF kernel. After optimization, the performance of SVM increases significantly.4. Raman spectra and SERS data of esophagus cancer tissues were selected for validation of SVM. The result shows that SVM has advantages over discriminating diagnosis.Our work indicated that, the classification of normal and abnormal samples by RBF kernel SVM algorithm performs well, and this research provides a new kind of classification method for Raman spectra data of caner.
Keywords/Search Tags:cancer, Raman spectroscopy, Surface-enhanced Raman spectroscopy, Support Vector Machine, RBF kernel, Parameters optimization
PDF Full Text Request
Related items