Font Size: a A A

Label-free Stages Classification Of Nasopharyngeal Carinoma Tissue Based On Raman Spectroscopy

Posted on:2018-09-08Degree:MasterType:Thesis
Country:ChinaCandidate:M Y LiuFull Text:PDF
GTID:2370330575973697Subject:Physical Electronics
Abstract/Summary:PDF Full Text Request
The nasopharyngeal carcinoma(NPC)detection was important to its prevention and treatment.The technique of Raman spectroscopy(RS)was employed to tumor tissues detection successfully.The deep learning algorithm as a popular data analysis method was widely applied to text analysis,voice technology and image identification.This study aims to suggest a electable method based on Raman spectroscopy for label-free nasopharyngeal carcinoma tissue staging.Meanwhile,the found chemical elements were deduced by analyzing classified mathematical process.The NPC tissues were obtained from two groups:30 early stage(?-?)NPC patients and 46 advanced stage(?-?)NPC patients.The molecular level information of the NPC tissue was detected by the Raman spectra technique.The early(?-?)and advanced(?-?)NPC tissues had different quantity and type of the biomolecule.It was revealed by the Raman spectra.In previous study,the Raman spectrum of the early(?-?)and advanced(?-?)NPC tissues were classified by linear discriminant analysis.It yielded classification sensitivities of 70%(21/30),specificities of 78%(36/46)and accuracy of 75%(57/76)by the leave-one-out cross validation.The deep learning algorithm as a nonlinear method was employed to analysis the Raman spectrum of the NPC tissues.It raised classification performance to sensitivities of 80%(24/30),specificities of 85%(39/46)and accuracy of 83%(63/76)by the leave-one-out cross validation.Raman spectroscopy technique as a non-invasive method combined with multivariate statistical analysis technique was employed for NPC tissues staging.This study made an effort to clinical practice of this method.According to the data analysis process,the significant peaks(1233 cm-1,1400 cm-1,1640 cm-1)of Raman spectra were found out.The different peaks of the Raman spectra corresponded to the certain chemical compositions(CH3/CH2 twisting,C-H vibration,Amide I)?Therefore,these significant chemical compositions for NPC staging were speculated.The result had significance for NPC patients treating correctly.
Keywords/Search Tags:nasopharyngeal carcinoma, staging, Raman spectroscopy, deep learning, stacked autoencoder
PDF Full Text Request
Related items