Shifted Excitation Raman Difference Spectroscopy for Identification of Oral Squamous Cell Carcinoma

Abstract: In recent decades, the research and development activities related to medical diagnosis have emphasized the development of real time, fast, and accurate methods of cancer screening. Raman spectroscopy has been typically proposed to replace conventional histopathological examination.

Identification of oral squamous cell carcinoma with a real time, portable and highly specific method is under investigation at the Lehrstuhl für Technische Thermodynamik (LTT) of the Friedrich-Alexander-Universität Erlangen-Nürmberg to increase the survival rate of patients of the cancer diseases by extracting its earlier information. Fluorescence rejection, which affects the overall accuracy of the useful information delivered by the Raman spectroscopy, is part of the research. The goal of this thesis is the development of a shifted excitation Raman difference spectroscopy (SERDS) technique combined with a baseline correction method which removes the overwhelming effect of fluorescence without losing relevant and significant Raman spectral features which hold the molecular profile.  It is also the aim of this thesis to classify tissues based on the reconstructed pure Raman spectra with a minimum misclassification errror. 

The overall work of this thesis is organized in three main parts. The first part of the work reviews Raman spectroscopy, its challenges and the available methods developed to tackle them. The second part is the implementation of a reconstruction algorithm of Raman spectra from a fluorescence corrupted spectra. A shifted excitation principle is applied to get rid of the fluorescence followed by baseline correction method to completely remove the fluorescence. The reconstruction method was able to get rid of the fluorescence without losing significant Raman information. The final part of the thesis is devoted to the results and discussions. Effect of excitation wavelength shift and strength of fluorescence are analyzed. Nine different tissues from four different pigs are differentiated with a misclassification error of 1.74%. Tissues of soft bone and hard bone showed the highest misclassifications. Significant differences, which are attributed to the spectral feature of lipids and proteins, among the Raman spectra of normal tissue and tumor tissue of the oral cavity of human are also observed.